Representations of the world: Memories,

perceptions, beliefs, intentions, and plans

PAUL KOCKELMAN

Abstract
Mental states are retheorized from the standpoint of social statuses (qua
commitments and entitlements to signify and interpret in particular ways)
and speech acts (qua signs with propositional contents). Using ideas developed in The semiotic stance (2005a), it theorizes five interrelated semiotic
processes that are usually understood in a psychological idiom: memories,
perceptions, beliefs, intentions, and plans. It uses this theory to account
for the key features of human-specific modes of intentionality (or theory
of mind), as well as the key dimensions along which culture-specific
modes of intentionality may vary (or ethnopsychologies). And it theorizes
emotion in terms of a framework that bridges the distinction between
social constructions and natural kinds.

1. Introduction
Intentionality is usually understood to be that quality of mental states
whereby they are directed at objects or states of aairs. For example,
Brentano, one of the first to theorize this quality, thought that each mental state includes an object within itself (1995 [1874]: 88), but not necessarily corresponding to something existing outside of the mind. In his
own words, [i]n presentation something is presented, in judgment something is armed or denied, in love loved, in hate hated, in desire desired
and so on (1995 [1874]: 88). Famously, Brentanos notion of intentionality was a major influence on Frege and Husserl; Frege and Husserl were
major influences on Wittgenstein and Heidegger, respectively; and Wittgenstein and Heidegger were major influences on analytic and continental
philosophy more generally (see Dummett 1994). In short, many of the
most important categories and cleavages within modern philosophy can
be traced back to intentionality in one guise or another.
Semiotica 1621/4 (2006), 73125
DOI 10.1515/SEM.2006.074

00371998/06/01620073
6 Walter de Gruyter

74

P. Kockelman

Modern scholars usually take Brentanos object within itself to be some

kind of propositional content, which may represent some state of aairs.1
And the entities that exhibit intentionality (or express propositional contents more generally) are taken to be either psychological entities (aka
mental states) or linguistic entities (aka speech acts). Besides having propositional contents, speech acts and mental states have propositional
modes or ways of relating to the propositional contents in question.
Thus, just as one may assert, promise, and forgive in the case of speech
acts, one may hope, believe, and want in the case of mental states. (Note
then that the terms speech act and mental state are used to refer to either the propositional mode alone, or both the propositional mode and
content together.) While philosophers seem to agree that the propositional contents of mental states and speech acts are equivalent, there are
long-standing debates regarding which kind of intentionality the psychological kind or the linguistic kind is originary and which is derivative (so far as it is inherited from the kind that is originary).2 Indeed,
there are even degrees of derivativeness. For example, in the case of linguistic intentionality, the intentionality of written language might be taken
to be derivative of the intentionality of spoken language. And in the case
of psychological intentionality, the intentionality of intentions and perceptions might be taken to be derivative of the intentionality of beliefs.3
In addition to keeping distinct the notions of proposition and state of
aairs, propositional mode and propositional content, speech act and
mental state, and originary and derivative intentionality, there are a number of other key distinctions to make. First, keep distinct the notions of
intentionality (as just characterized) and intentions (as purposes with
propositional content, as characterized in Residence in the world). Intentions are just one species of intentionality, taking their place alongside
other species such as belief, desire, perception, and so forth. Second, intersecting the philosophical literature on intentionality is a more recent
literature on theory of mind and ethnopsychology, stemming from disciplines like psychology, primatology, and anthropology (see, for example,
Lillard 1998; Premack and Woodru 1978). If theory of mind refers to
cross-cultural ways of understanding others in terms of intentionality,
ethnopsychology refers to culture-specific ways of understanding others
in terms of intentionality (where, owing to the disciplines that take it up,
intentionality is usually understood in terms of mental states and not
speech acts). And third, keep distinct the capacity to have ones behavior
understood in term of intentionality (regardless of why or how one actually behaves), and the capacity to understand others behavior in terms of
intentionality (regardless of why or how they actually behave). For example, many of us might be inclined to understand the behavior of a rabbit

Representations of the world

75

(or robot) in an intentional idiom, and understand it predictably well

(say, in term of wanting carrots, fearing predators, envying the Easter
Bunny, and so forth). However, most of us do not expect a rabbit (or
robot) to understand others behavior, including the behavior of its conspecifics, in an intentional idiom Bugs Bunny (and C3PO) aside.
In this essay, the notion of an intentional status is put forth to replace
the notion of a mental state. In particular, an intentional status is a set of
commitments and entitlements to signify and interpret in particular ways:
normative ways of speaking and acting attendant upon being a certain
sort of person a believer that the earth is flat, a lover of dogs, one
who intends to become a card shark, and so forth. An intentional role
is any enactment of that intentional status: actually putting one or more
of those commitments and entitlements into eect; or speaking and acting
in a way that conforms with ones intentional status. And an attitude is
just anothers interpretant of ones status by way of having perceived
ones roles: I know you are afraid of dogs, as an intentional status, insofar as I have seen you act like someone afraid of dogs; and as a function
of this knowledge (of your status through your role), I come to expect
you to behave in certain ways and perhaps sanction your behavior as
a function of those expectations. In this way, an intentional status is
treated as a special kind of social status: one which involves a mode of
commitment (which is a semiotic way of accounting for propositional
modes, a` la belief and desire) and a content of commitment (which is a
semiotic way of accounting for propositional contents, a` la what is believed and what is desired). In short, so-called mental states are theorized in terms of social statuses, on the one hand, and speech acts, on the
other.
This terminology is introduced to reframe seemingly psychological
questions in a social and semiotic idiom. If assuming an intentional
stance involves using an intentional idiom to understand behavior (Dennett 1978), assuming a semiotic stance involves using a semiotic idiom to
understand behavior. This essay demonstrates how The semiotic stance
accounts for the intentional stance more generally.
Whatever the inclination of scholars (intentional status versus mental
state, semiotic stance versus intentional stance), everyone agrees that the
directedness of any entity (i.e., an intentional status, or semiotic process
more generally) is only meaningful in the context of other directed entities. As Haugeland puts it (1998: 130), at some level of description there
must be a non-accidental larger pattern within which the directedness of
any particular entity makes sense. In the case of non-propositional modes
of semiosis, the non-accidental larger pattern was called the residential
whole. In the case of propositional modes of semiosis, it may be called

76

P. Kockelman

the representational whole. If the key constituents of the residential whole

were aordances (who objects are purchases), instruments (functions),
actions (purposes), roles (statuses), and identities (values), the key constituents of the representational whole are intentional statuses loosely
correlated with what are usually called memories, perceptions, beliefs, intentions, and plans (whose emblematic roles are various kinds of speech
acts and non-verbal behaviors).4
Like the constituents of the residential whole, the constituents of the
representational whole are holistically interrelated by various modes of
coherence. In particular, each is inferentially articulated relative to other
constituents of the representational whole (being able to stand as a reason
or stand in need of a reason); and each is indexically articulated relative
to other constituents of the residential whole (being causal of, or caused
by, such constituents). For example, and loosely speaking: beliefs may
justify and be justified by other beliefs; perceptions may justify beliefs
and be caused by states of aairs; and intentions may be justified by beliefs and be causal of states of aairs. Residence in the world and representations of the world, taken together as irreducibly interrelated, provide
a social and semiotic theory of being-in-the-world.
Besides the large-scale move from the intentional stance to The semiotic stance, and from mental states to intentional statuses, there are a
number of general arguments put forth in this essay. First, as semiosis
must primarily be understood in terms of sign-interpret relations rather
than sign-object relations, intentionality should primarily be understood
in terms of inference rather than representation. In this regard, the account of inferentialism put forth in Brandom (1994) plays a large role in
section 5. And in the background of this essay are philosophers like Kant,
Frege, Wittgenstein, Sellars, and Dummett. Second, rather than coming
to inferentialism through analytic philosophy, this essay comes to it
through American pragmatism and Boasian anthropology. In this regard,
this essay continues with the model of semiosis that was developed in
The semiotic stance, and which was inspired by Peirce, Mead, Linton,
Goman, and linguistic anthropology more generally. Third, rather than
oer an account of propositionally contentful signs and either ignore nonpropositionally contentful signs, or state that they are important without
oering a complementary account, this essays builds on the essay entitled
Residence in the world. Fourth, rather than argue about intentionality
in ahistorical and/or species-specific terms, as philosophers tend to do,
this essay follows scholars like Vygotsky and Tomasello in keeping distinct interactions among processes that occur on three dierent timescales: phylogenetic (species-level mediation), historical (culture-level mediation), and ontogenetic (individual-level mediation).5 And finally, the

Representations of the world

77

accounts of residence in the world and representations of the world need

to be related to agency and selfhood more generally, as discussed in
Agent, person, subject, self.
With these assumptions and stakes in mind, the theory of intentionality
to be presented in this essay may be understood as seven modes of thirdness in which any intentional status is implicated.6 Section 1 shows how
an intentional status may be understood as an ultimate (representational)
interpretant: any number of dierent kinds of sign events can lead to it
(constituting its roots) or follow from it (constituting its fruits). Part of
what it means, then, to understand the behavior of others as intentional,
to read others minds, is to be able to infer fruits from roots, by reference to the intentional statuses that mediate between them (and vice
versa).
Section 2 discusses the phylogenetic, historical, and ontogenetic mediation of intentional statuses. Using the work of Michael Tomasello
and colleagues, it outlines the cultural and cognitive conditions for,
and consequences of, this mind-reading or rather sign-interpreting
ability.
Section 3 shows how an intentional status may be understood as a semiotic object: any number of intentional roles can stand for it (as signs);
and any number of intentional attitudes can interpret it. And it theorizes
relatively emblematic roles of intentional statuses: modes of behavior,
such as facial expressions and speech acts, which provide relatively incontrovertible evidence of ones intentional statuses.
Section 4 discusses the grammatical properties of utterances used to ascribe intentional statuses, and frames these in terms of their iconic, indexical, and symbolic properties. It argues that (such relatively emblematic
roles of ) intentional statuses may be cross-linguistically ordered and categorized as a function of the ontological disjuncture between the intentional status and the state of aairs it represents: for the example, the extent to which an event of believing or desiring is causally implicated in the
event believed or desired.
Section 5 shows the inferential and indexical articulation of intentional
statuses with propositional content. And it oers a detailed account of the
five most fundamental intentional statuses: memories, perceptions, beliefs, intentions, and plans.
Section 6 shows how modes of commitment acquire propositional content through intentional status predicates (ISPs) that refer to them (e.g.,
words like believe, perceive and intend); and it details how these ISPs
allow speakers both to predicate intentional statuses of people (e.g., John
believes shell go) and to predicate properties of intentional statuses (e.g.,
belief is a weak form of knowledge).

78

P. Kockelman

Section 7 shows how intentional statuses become implicated in epistemes that allow them to be the objects of empirical investigations, theoretical representations, and practical interventions.
And section 8 shows how the phenomena currently grouped under the
term emotion can be understood relative to the foregoing framework.

2.

Antecedent sign events, consequent sign events, and mediating

embodied signs

In The semiotic stance, it was argued that many interpretants are ultimate interpretants: they involve a change in status, where status is understood as a set of commitments and entitlements to signify and interpret in
particular ways (as evinced in the sanctioning practices of a community
and as embodied in the dispositions of its members). Unlike most interpretants, such ultimate interpretants are not signs that stand for something else; rather, they are dispositions to signify and interpret. And insofar as they are not signs that stand for something else, but merely
dispositions to signify and interpret, they are invisible being known
only through the signs that lead to them (insofar as they are interpretants), or through the patterned modes of signifying and interpreting that
follow from them (insofar as they are dispositions to signify and interpret).7 Using Colapietros metaphor (1989), the underlying idea is that ultimate interpretants have both roots and fruits.
These ultimate interpretants were also called embodied signs to stress
that they involved thirdness (like signs more generally), but that they
were non-sensible (unlike signs more generally). Examples were oered
such as social statuses (being a mother or banker) and intentional statuses
(believing god is dead or intending to grow a large mustache).8 And the
idea of semiotic framing was put forth: on the one hand, sign events may
be understood to lead to, and follow from, embodied signs; and on the
other hand, embodied signs may be understood to lead to, and follow
from, sign events. That is, one can focus on relatively public sign events
(such as speech acts) that lead to or follow from relative private embodied
signs (such as social and intentional statuses); or, inverting the frame, one
can focus on relatively private embodied signs that lead to or follow from
relatively public sign events.
The two views are equivalent, like the two perspectives of a Necker
cube, but the former focuses on non-sensible entities being mediated
through sensible events; and the latter focuses on sensible events being
mediated through non-sensible entities. It may now be seen that semiotic
framing allows one to focus on intentional statuses (the non-sensible

Representations of the world

79

entities) or on speech acts (the sensible events) or, generalizing to nonpropositional semiosis, semiotic framing allows one to focus on either embodied signs or sign events. In short, at this level of analysis, the distinction between derivative and originary intentionality is reduced to a
question of semiotic framing, such that to ask which came first the
sign event or the embodied sign (the speech act or the mental state) is
equivalent to asking whether it was the chicken or the egg.
The basic structure of these embodied signs, with their roots and fruits,
is shown in figure 1. There is a class of antecedent sign events, or roots
(labeled A1, A2, A3, etc.). This is the class of signs events that lead to
the embodied sign (so far as it is an interpretant of them). There is a class
of consequent sign events, or fruits (labeled C1, C2, C3, etc.). This is the
class of sign (and interpretant) events that follow from the embodied sign
(so far as it disposes one to signify (and interpret) in particular ways).
And there is the non-sensible but inferable mediating variable that links
them (M). This is the embodied sign (or intentional status) itself.
To summarize the peculiar features of embodied signs, note the following. The antecedent and consequent sign events are sensible; whereas the mediating variable is non-sensible. The antecedent sign events
are prior to the consequent sign events (as presupposed by the terminology). There is no mapping between particular antecedent sign events
and particular consequent sign events; there is only a mapping between
the class of antecedent sign events and the class of consequent sign
events. That is, any antecedent sign event can lead to the mediating
variable, and any consequent sign event can follow from the mediating
variable (i.e., following either the seeing of a rattlesnake or the hearing
of ssst could be either the yelling of snake or the unholstering of a pistol). The class of antecedent sign events, like the class of consequent
sign events, is heterogeneous: the various antecedent sign events, and
the various consequent sign events, are not sensible tokens of a common type (i.e., seeing a rattlesnake has very little in common, as a phenomenological experience, with hearing ssst).9 Their commonality as a
class is entirely due to their leading to, or following from, the same
mediating variable. Thus, to define antecedent sign events as a class,
one must necessarily specify the consequent sign events as a class, and
the mediating variable between them (and vice versa). There need not
be a many-to-many relation between the number of dierent antecedent
events and the number of dierent consequent events: there may be a
many-to-one relation, a one-to-many relation, a few-to-few relation, a
few-to-one relation, and so forth. Finally, temporal latency is possible.
That is, while the consequent event must follow the antecedent event, it
need not follow it instantaneously.10

80
P. Kockelman

Figure 1. Antecedent sign events, consequent sign events, and mediating embodied sign. See Tomasello and Call (1997: 37).

Representations of the world

81

Note how the mediation of antecedent and consequent sign events

diers from the pairing of stimuli and responses in the behavioralist
idiom. In particular, the class of stimuli, like the class of responses, consists of sensible tokens of a common type. Any response must immediately follow any stimulus. And there is no need to postulate a mediating
variable: the stimuli and the responses are all there is. For these reasons,
the triadic structure of embodied signs begins to look like the dyadic
structure of stimulus-response pairing only in the limit that all antecedent
sign events are sensibly alike and all consequent sign events are sensibly
alike; only in the limit that the consequent sign events immediately follow
the antecedent sign events; and only in the limit that no embodied sign (or
mediating variable) is required to explain the relation.
These embodied signs also known as ultimate (representational) interpretants, social and intentional statuses, and/or mediating variables
are worth exemplifying in detail. They may be thought of as emotions
(in the pre-theoretical sense that will be disavowed in section 8). For example, jealousy as an embodied sign can follow from any number of antecedent sign events (seeing your spouse being intimate with a stranger,
reading about your lovers fantasy in his diary), and can lead to any number of consequent sign events (making an appointment with ones analyst,
attacking the stranger, berating ones spouse, crying uncontrollably). And
anger as an embodied sign can follow from any number of antecedent
sign events (reading the headlines, not sleeping, spilling soup), and can
lead to any number of consequent sign events (becoming irritated by car
alarms, being brusque with the questions of children, taking personal
oence at grati). As theorized in the Residence in the world, embodied signs may also be thought of as the objects of the constituents
of the residential whole: purposes, statuses, and values.11 Indeed, in the
spirit of Aristotles final cause, William James (1950 [1890]: 68) had a
prescient understanding of embodied signs that had been cut in half:
any number of means may lead one to infer an identical end. Embellishing his famous example, a desire or intention to woo Juliet (as an
embodied sign) leads to a relatively heterogeneous class of consequent
sign events (ringing the doorbell, knocking on the door, climbing the
wall, sending a telegram, waiting for her to come outside, practicing the
pole vault, and so on), such that if one of them is frustrated Romeo can
try another, and such that an observer of Romeos actions can explain his
many disparate behaviors with a single purpose. And, as will be seen, perceptions, intentions, wishes, memories, beliefs, and plans have very similar structures: any number of sign events can lead to the same intentional
status; and any number of sign events can follow from the same intentional status.

82

P. Kockelman

The antecedent and consequent sign events of embodied signs may be

other embodied signs. This leads to patterns that are reminiscent of the
normative pairings between circumstances and behaviors examined in
The semiotic stance (for the simple reason that dispositions are the
most common kinds of embodied signs). There is chaining: an antecedent
sign event leads to an embodied sign which leads to a consequent sign
event which leads to another embodied sign (as its antecedent sign event).
There is nesting: an antecedent sign event leads to an embodied sign
which leads to another embodied sign which leads to a consequent sign
event. There is feedback: an antecedent sign event leads to an embodied
sign which leads to a consequent sign event which leads to the same embodied sign (as its antecedent event). There is conditional mediation: an
antecedent sign event leads to an embodied sign only in the context of
there already being another embodied sign (and hence another antecedent
event). And so on, and so forth. Indeed, one sense of thinking or semiotic action or interiority is really the idea that embodied signs can beget
other embodied signs without the necessary mediation of (publicly sensible) antecedent or consequent sign events. Needless to say, these types of
patterns make behavior exceptionally complicated, and make interpreting
behavior necessarily holistic.
With these examples in mind, it should be noted that there is a major
dierence between an entity that understands event-sequences in terms of
stimulus-response pairings, or secondness, and an entity that understands
them in terms of mediating embodied signs, or thirdness. As a mathematical analogy, imagine being given a set of numerical correlations: 1 and 1,
2 and 4, 3 and 9, 4 and 16, and so forth. A non-mathematically inclined
person may remember the set of individual pairings, and hence come to
expect quite accurately that 1 goes with 1, 4 goes with 2, 9 goes
with 3, and so forth. In contrast, a mathematically inclined person
one who can infer the mediating function, y x 2 , from the set of pairings
is able to predict pairings she has never seen: 25 and 5, 36 and 6, and
so forth. The dierence between a beast that can note correlations between individual variables, and a beast that can infer functions between
types of variables, is analogous to the dierence between a beast that
can note correlations between individual antecedent events and individual
consequent events, and a beast that can infer mediating variables between
classes of antecedent events and classes of consequent events. In short, as
the non-mathematically inclined is to the mathematically inclined in the
realm of variables and functions, second-seeing entities (or behavioralists) are to third-seeing entities (or semioticians) in the realm of sign
events and embodied signs. And one suspects that there are dierences in
individuals, not just dierences in species in the realm of mathematics

Representations of the world

83

and physical causes there are Newtons, and in the realm of intentionality
and psychological motivations there are Dostoyevskys.

3. Embodied signs in phylogenetic, historical, and ontogenetic timescales

In the literature on primate cognition, something empirically (but not theoretically) analogous to what are here called embodied signs, have been
called mediating variables (Whiten 1993) and/or tertiary relations
(Tomasello and Call 1997: 383). While the ideas of these scholars are
couched in terms of Aristotles final cause rather than sociality and semiosis, it is useful to review their general arguments.12
The primatologists Tomasello and Call (1997) note that intentionality
and causality both involve temporally ordered events (antecedentconsequent relations), where the antecedent event and the consequent
event are external to the observer, and where there is some inferred
intermediary cause or goal that organizes and explains the event sequence such that dierent antecedents may lead to the same consequent
. . . and the same antecedent may lead to dierent consequents in dierent
circumstances (Tomasello and Call: 1997). Their crucial argument is this:
while a non-human primate may understand the antecedent-consequent
relations among external events in the absence of its own involvement,
it does not understand the mediating forces in these external events that
explain why a particular antecedent-consequent sequence occurs as it
does and these mediating forces are typically not readily observable
(Tomasello 1999: 23). That is, non-human primates cannot infer mediating variables: they cannot understand embodied signs.
In this regard, Tomasello and Call (1997) argue that understanding the
relevant intermediaries in a particular case enables individuals to devise
novel ways of producing the intermediary and thus the end result. That
is, in a particular circumstance in which the usual antecedent event is not
present or not possible, an organism should in some cases be able to create a dierent antecedent event leading to the same intermediary and thus
to the same result as usual (e.g., creating fear in conspecifics so that they
will scatter or creating the movement of a limb so that fruit will fall)
(1997: 390). For example, and in the idiom introduced here, when a purpose is the embodied sign at issue, one can assess it independent of the
means used (or controlled behavior undertaken) to realize it, and even
independent of its realization (say, in cases of frustrated action). Notwithstanding their inability to infer mediating variables (or understand embodied signs), non-human primate behavior is still exceptionally rich and
complicated. Tomasello and Call (1997: 388) argue that this is due to

84

P. Kockelman

several interrelated capacities: one, non-human primates understand

conspecifics as animate and directed, and hence understand stimulusresponse pairs between environmental events and behavioral events; and
two, they understand tertiary social relations (as discussed in section 1 of
The semiotic stance): not just pecking orders (how ego relates to alter),
but egos ability to infer how one alter relates to another alter, insofar as
ego knows how it is related to each of these alters. The complicated behavior enabled by such capacities often looks like an ability to interpret
embodied signs; hence, all the anecdotes of monkey mind reading.
While subsequent research has shown that this hypothesized cognitive
distinction between human primates and non-human primates is not so
clear cut (see Tomasello, Call, and Hare 2003, and references therein), it
is still a major distinction. Phrased in terms of embodied signs, and in a
more tempered fashion, non-human primates can understand the behavior of conspecifics in terms of pair-wise correspondences between individual antecedent events and individual consequent events; but they are not
particularly good at seeing the mediating relation between the class of antecedent events and the class of consequent events. In short, in terms of
understanding the behavior of their conspecifics, non-human primates are
much better behavioralists than they are semioticians.
As noted in the introduction, and as just emphasized, it is important to
distinguish between the capacity to have ones behavior understood in an
intentional idiom, and the capacity to understand anothers behavior in
an intentional idiom. In this regard, it is useful to provide a hierarchy of
dierent ways an entity may be understood as understanding the behavior
of other entities. The following hierarchy is inspired in part by Tomasello
and Call (1997: 190), and in part by Brandom (1994: 46) and Sellars
(1997 [1956]) though its categories are taken from The semiotic
stance. To understand an entity as responsive is to understand it as reacting predictably to changes in its environment: some class of causes
leads to some class of eects. A rock, for example, may be understood
as responsive (to gravity, wind, and so forth). To understand an entity as
animate is to understand it as generating its own behavior, or as selfmoving as Aristotle called it. To be animate, in a sense, is to be selfresponsive. To understand an entity as animate, then, is to not be surprised when it moves on its own accord. To understand an entity as
sentient is to understand it as both predictably responding to sensible
events with behavioral events and predictably causing sensible events
with its behavioral events. That is, one expects one-to-one causal relations
from the entity to the environment and from the environment to the entity (where the environment can be other such entities): one is not surprised that certain circumstances can lead to changes in its behavior; nor

Representations of the world

85

surprised that certain of its behaviors can lead to changes in circumstance. This is the realm of behavioralism. To understand an entity as
semiotic is to understand it as having embodied signs which mediate between antecedent sign events and consequent sign events, and hence interpreting sensible events (or antecedent sign events) via embodied signs,
and as interpreting embodied signs via sensible events (or consequent
sign events). Indeed, there is nothing anthropomorphizing or fetishizing in the fact that humans see animals as having purposes. Rather, animals experience purchasefully and behave purposefully and any human account of animal behavior must turn on this (even if animals dont
understand other animals behavior as such and hence dont purposively act to change the purposes of other animals). Finally, to understand an entity as sapient, is to understand the embodied signs that organize its behavior as having propositional content, and hence being both
indexically related to the world and inferentially related to other embodied signs. This is the realm of the representational whole, and the exemplary embodied signs are perceptions, beliefs, and intentions.
Using this idiom, one can contrast human primates and non-human
primates, and infant and non-infant human primates. In particular, nonhuman primates are relatively responsive, animate, sentient, and semiotic;
but they only understand other conspecifics as responsive, animate, and
sentient. Human primates are relatively responsive, animate, sentient, semiotic, and sapient, but only begin to understand other conspecifics as semiotic between nine and twelve months (as evinced in joint-attention),
and only begin to understand other conspecifics as sapient between one
to three years (as evinced in language use).13 Human primates also, of
course, tend to understand non-human primates and infant humans as responsive, animate, sentient, semiotic (correctly), and sapient (incorrectly).
Indeed, the ethnographic record shows that human primates are willing
to understand just about every process, natural or cultural, as sapient.14
And one suspects that, just as humans anthropomorphize non-human primate behavior (treat sentient and semiotic behavior as sapient), nonhuman primates probably simian-pomorphize human primate behavior
(treat sapient and semiotic behavior as sentient). And much of Heideggers critique of western metaphysics is that it attempts to understand
semiosis in terms of sapience, or non-propositional modes of semiosis in
terms of propositional modes of semiosis: to understand residence in the
world in terms of representations of the world. Attempts to understand
the behavior of entities in terms of capacities that are above or below its
understanding thereby bumping it up or down this hierarchy are
part and parcel of what anthropologists call fetishization and reification, respectively.

86

P. Kockelman

Tomasello (1999) has argued that the human capacity to see other conspecifics as semiotic (what he calls intentional) is a phylogenetic adaptation, and the ability to see other conspecifics as sapient (what he calls
mental) is the precipitate of this phylogenetic adaptation in conjunction
with ontogenetic development in historical time. His hypothesis is quite
elegant, and it should be paraphrased in terms of the framework oered
here (1999: 10, 48). Human primates have cognitive skills that originated
via biological inheritance working in phylogenetic time in particular,
the ability to identify with conspecifics and thereby understand them in
semiotic (or intentional) terms. That is, human primates can understand
the embodied signs of other human primates (in particular, their purposes). With these cognitive skills, and the modes of socialization and
sociogenesis they allow, they exploit cultural resources that evolved in
historical time which we might theorize as all the culture-specific constituents of the residential whole (aordances, instruments, actions, roles,
and identities) and all the culture-specific constituents of the representational whole (perceptions, memories, beliefs, intentions, and plans). And
they do this in ontogenetic time. In particular, benefiting from accumulated historical traditions (i.e., the constituents of the residential and
representational wholes), through joint-attention they learn linguistic
symbols (and subsequently all the language-specific cognitive resources
that such symbols enable: construal, metaphor, displacement, generativity, performativity, etc.), and they come to internalize complicated constructions involving these symbols (and hence acquire dialogic thinking,
meta-cognition, and related discourse-based cognitive resources). Broadly
speaking, then, it is a theory of intentionality that makes reference to
complicated interactions among dierent processes that are occurring on
three separate time-scales (phylogenetic, historical, ontogenetic), and
thereby complicates many of the more rarified philosophical arguments
regarding primary and derivative intentionality.

4.

Intentional statuses, intentional roles, and intentional attitudes

The last two sections focused on embodied signs: the invisible forces that
mediate between antecedent sign events and consequent sign events. Using Lintons understanding of status and Peirces understanding of ultimate (representational) interpretants, they were understood both as a
complex kind of interpretant that an antecedent sign event could lead to,
and as a complex kind of sign that a consequent sign event could follow
from. Indeed, through semiotic framing, they could be understood not
only as (embodied) signs, but also as (dynamic) objects and (ultimate)

Representations of the world

87

interpretants. Those embodied signs that seemed the most psychological

were gathered together under the term intentional status. The basic idea
behind roots and fruits, then, is that intentional statuses have all the properties of signs except that they are embodied and private rather than expressed and public. In this section, another related mode of thirdness is
taken up: how expressed and public signs stand for these intentional statuses, and give rise to interpretants which may themselves be other intentional statuses. In short, having examined intentional statuses, one may
now examine the intentional roles that signify them, and the intentional
attitudes that interpret them.
The first thing to notice is that there is no isomorphism between intentional roles and intentional statuses, or between signs and objects. In particular, under the right semiotic frame, the same role (or sign more generally) can stand for many dierent intentional statuses (or objects more
generally), and the same intentional status (or object more generally) can
be stood for by many dierent roles (or signs more generally). For example, an infinite number of behaviors and utterances could be said to index
disgust as its roots or as its fruits: a facial expression, an interjection
like yuck or gross, turning up ones nose, saying Im disgusted, leaving
a room in a hurry, never standing downwind of a particular establishment, never entering a particular part of town, obsessively washing ones
hands, using large amounts of perfume, stumbling upon fresh feces, nervously fingering a barf bag, and so on. Indeed, any one of these behaviors
or utterances could also index any number of other objects for example, social statuses like gender, age, class, nationality, and so forth.
As an example in the other direction, one may turn to Qeqchi-Maya,
a language spoken in Guatemala (see Kockelman 2003a for a full discussion). Here the interjection chix is commonly thought to index ones disgust (as an intentional status). However, chix also indexes a number of
other objects. For example, it may index an object that causes the disgust
(e.g., chicken feces, vomit, rotten eggs, and so on), and thereby serve to
call anothers attention to the oending object. It may index a sign of
such a disgusting object (e.g., as another is describing the carcass of a
dead dog they found in the river), thereby functioning as a back-channel
cue, indexing the speakers interest in what their interlocutor had just
been saying. In such a usage, it often indexes the gender of the speaker
as well, being said by women in the context of back-channeling, and
thereby signaling that the male speaker should continue speaking. And it
may index anothers proximity to an object of disgust, and thereby serve
as an imperative not to touch the object. In such a usage, it may index the
status of the speaker as well, signaling that they are in a care-giver relation to the one who is in proximity to the object of disgust (e.g., a parent

88

P. Kockelman

Table 1. The four dimensions of relatively emblematic roles

Phenomenological

Relational

Normative

Epistemic

A role which is
maximally public
(i.e., perceivable
and interpretable);
and a role which
is minimally
ambiguous (i.e.,
one-to-one and
onto).

A role which all

members of an
intentional status
have in common;
a role by which
members of
dierent
intentional
statuses contrast;
and a role of
which all members
are conscious.

A role which may

(only) be
expressed by
members of a
particular
intentional status;
and a role which
must (always) be
expressed by
members of a
particular
intentional status.

A role which
provides necessary
and sucient
criteria for
inferring (and/or
ascribing) the
intentional status
in question.

in relation to a child who is about to touch something gross). In short,

any behavior or utterance can index an intentional status, and thereby
be an intentional role; and any behavior or utterance that seems to index
an intentional status, and thereby be an intentional role, can index objects
other than the intentional status.
With this lack-of-isomorphism in mind, the following typology of common intentional roles (or signs of intentional statuses) may be developed.15 The point behind this typology is to begin examining relatively
emblematic intentional roles. That is, 1) roles which are minimally ambiguous and maximally public; 2) roles which members of an intentional status have in common, by which members of dierent intentional statuses
contrast, and of which all such members are conscious; 3) roles which
may only and must always be expressed by members of a particular intentional status; 4) roles which provide necessary and sucient evidence for
inferring or ascribing the intentional status in question. See table 1. Intentional roles which satisfy all of these criteria are maximally emblematic;
however, anyone of these criteria, and often several at once, may be relaxed, and still a role is relatively emblematic. Relatively emblematic intentional roles, then, are the semiotic resources we have for dealing with
this otherwise lack-of-isomorphism between intentional roles and intentional statuses. They constitute relatively incontrovertible evidence of our
intentional statuses.
First, there are signs of modes of commitment (a` la propositional
modes or intentional states) and signs of contents of commitment (a` la
propositional contents or intentional objects). For example, there are
signs that indicate the state of aairs that ones intentional state is directed at (say, that lions are dangerous or that he is a lawyer); and

Representations of the world

89

there are signs that indicate the type of intentional state that is directed
(say, believe or hope).
Second, these signs of modes of commitment and contents of commitment may have propositional contents or not. For example, I feel disgusted by the mess in your room is a relatively propositional sign of
both the content of commitment (the mess in the room) and the mode
of commitment (the feeling of disgust). In contrast, yuck is a nonpropositional sign of the content of commitment (e.g., the disgusting object it indexes), and a non-propositional sign of the mode of commitment
(e.g., disgust per se).
Third, propositional signs of modes of commitment or contents of
commitment may involve grammatical forms (or operators) or lexical
forms (or predicates). For example, the modal auxiliary verbs may and
must are grammatical operators that can indicate degrees of deontic (and
epistemic) commitment and entitlement; whereas the adjectives permitted
and obligated are lexical predicates that can indicate degrees of deontic
entitlement and commitment. Other grammatical operators include verbal categories such as status (e.g., I was going to go), mood (e.g., take
out the trash), and illocutionary force (e.g., if only it would rain). And
other lexical predicates include what linguists call propositional attitudes
denoting intentional statuses words like believe, know, and hope.
Fourth, non-propositional signs of modes of commitment or contents
of commitment may involve indexical-symbols or iconic-indices. For example, interjections like ugh (and many forms of prosody) are relatively
indexical-symbols of modes of commitment, and relatively iconic-indices
of contents of commitment. And facial expressions (as well as tears and
laughter) and direction of attention (as well as changes in it, as occur in
joint-attention) are relatively iconic-indices (and sometimes indexicalsymbols) of modes of commitment and iconic-indices of contents of commitment. (Notice that the iconic-index/indexical-symbol division is not
the same as the grammatical operator/lexical predicate distinction. And
note that both are relative notions.)16
This typology correlates with several other important semiotic and social features. First, as a general tendency, the more an intentional status
has a sign of both its mode and content of commitment, the more a sign
is propositional (versus non-propositional), the more a sign is lexical
(versus grammatical), and the more a sign is indexical-symbolic (versus
iconic-indexical), the more the signer can control the expression of its
sign, the more the signer can composes the sign-object relation, and the
more the signer can commit to the interpretant of the sign-object relation.
Insofar as the question of control, composition, and commitment arises
with any semiotic process, one can ask these questions of intentionality.

90

P. Kockelman

In particular, to control the expression of a sign is to control the expression of an intentional role; to compose the relation between a sign and a
object is to compose the relation between an intentional role and an intentional status; and to commit to the interpretant of a sign-object relation is to commit to the intentional attitude of an intentional role-status
relation. For example, one has fewer degrees of control, composition,
and commitment with facial expression than with interjections, with
modal auxiliary verbs than with propositional attitudes, and so forth. Indeed, partially scaling with this typology is a distinction between normative regimentation and causal regimentation: hence, facial expressions will
tend to be more cross-cultural than interjections; and grammatical operators will tend to be more cross-linguistic than propositional attitudes. In
short, this typology of common pairings of intentional statuses with intentional roles correlates with the degree of sharedness (across semiotic
communities) and the degree of sincerity (between expressing of role and
inhabiting of status) of the pair.

5.

Iconic, indexical, and symbolic properties of grammatical patterns

involving Intentional Status Predicates

While explicit reference to grammatical patterns has been avoided in

these essays, one key grammatical pattern involved in propositional signs
of modes of commitment and contents of commitment (or intentional statuses more generally) should be mentioned: the notion of tightness. This is
a dicult idea, but it is worthwhile to go grammatically out of the way to
discuss it insofar as the linguistic constructions which express it are also
the linguistic constructions that maximize the degree of control, composition, and commitment, insofar as both the mode and content of commitment are signified by signs with propositional content. Indeed, many of
the most emblematic intentional roles (in particular, speech acts, such as
I believe that it will rain), may have their grammatical structure characterized in terms of tightness.
It is well known that across languages the semantic structure of
complement-taking predicates (or verbs that have as their arguments constructions involving other verbs) correlates in a relatively systematic way
with the morphosyntactic structure of their complements (or the constructions involving the other verbs) (cf. Givon 1980; Kockelman 2003b,
2005b; Silverstein 1993; Van Valin and LaPolla 1997). In particular, the
closer the semantic relation between the narrated events denoted by a
predicate and its complement, the more the morphosyntactic encoding of
the predicate-complement construction appears as a single clause. Given

Representations of the world

91

that modes of commitment are often denoted by complement-taking

predicates (e.g., I believe or I want), and given that contents of commitment are often denoted by the complements of these predicates (e.g., that
he is a lawyer, or to go to the store), this shows that propositional signs of
intentional statuses may be grouped and ordered as a function of their
tightness: the degree to which the mode of commitment and the content
of commitment express a single event, or the degree to which the
complement-taking predicate and the complement are expressed in a single clause.
This iconicity may be demonstrated in English by comparing two signs
of modes of commitment events: full-clause constructions involving the
verb believe and non-finite constructions involving the verb want. An example of a full-clause complement in English is the clause following the
complementizer that in the sentence John believes that Mary might have
been a witch. Notice that in full-clause constructions the subject of the
complement verb may be dierent from that of the main verb, and the
timing and truth-value of the action denoted by the complement verb
may be dierent from that of the action denoted by the main verb: John
is the subject of believe while Mary is the subject of be; and Johns belief
is true at the time of the utterance, while Marys being a witch is possibly
true before the time of the utterance. In other words, the grammatical
encoding of these two events (each can be distinctly inflected for personnumber and tense-aspect-modality) resembles the semantic relation denoted by this encoding (Marys being a witch is a relatively distinct event
from Johns believing it).
An example of a non-finite complement in English is the verb be in the
sentence John wants to be a warlock. Notice that, unlike in full-clause
constructions, the subject of the complement verb is the same as that of
the main verb, and the timing and truth-value of the action denoted by
the complement verb is directly related to that of the action denoted by
the main verb: John is the subject of both want and be; and Johns being
a warlock is constrained by the timing and truth-value of Johns desire.
Again, the grammatical encoding of these two events (the complement
cannot be distinctly inflected for person-number and tense-aspectmodality) resembles the semantic relation denoted by this encoding
(Johns becoming a warlock is directly related to his desire to be a
warlock).
As the event of wanting (as a mode of commitment) is more closely implicated in the event wanted (as a content of commitment) than the event
of believing (as a mode of commitment) is implicated in the event believed (as a content of commitment), the predicate-complement constructions in which the verb want are implicated are more like a single clause

92

P. Kockelman

than the predicate-complement constructions in which the verb believe are

implicated. This form-functional iconicity presents a range of relative
tightness along which complement-taking predicates can be scaled relative to one another. In this example, predicate-complement constructions
involving want are tighter than those involving believe: their morphosyntactic encoding looks more like a single clause, and the events encoded
look more like a single event. This is the logic underlying the lexical expression of intentional statuses.
Givon (1980) has noted that English complement-taking predicates
may be ordered as follows (moving from less tight to more tight constructions): say and tell; think, know, believe, doubt, and learn; decide and
agree; like, hope, expect, love, and hate; plan, intend, and try. This list
moves from predicates that may take full clauses, to predicates that may
take both full clauses and infinitives, to predicates that may only take infinitives. And Kockelman (2003b, 2005b) demonstrates this hierarchy in
Qeqchi, a Mayan language, and relates it to the function of grammatical
categories like status (which marks features such as facticity, counterfactivity, afacticity, and so forth). This form-functional iconicity may also be
generalized across languages (Van Valin and LaPolla 1997: 478479) using the following hierarchy of potential constructions (ranging from denotata that are most like two events to denotata that are most like one
event): unrelated events; sequential events; simultaneous events; conditionals (if-then constructions); reported speech (say); cognition (know,
think); propositional attitude (believe, consider); perception (see, hear);
jussive (ask, order); purposive (go, come); psych-action (forget, want); aspectual (start, continue); and causative (let go, push open). Regarding the
intentional statuses at issue in this essay, the following ordering generally
holds (from least tight to most tight): belief and knowledge, memory, perception, planning, intention.
If these complement-taking predicates are understood to denote modes
of commitment in particular, cognition predicates, propositional attitude predicates, perception predicates, and psych-action predicates
then the logic which orders intentional statuses may be seen: the greater
the tightness between a mode of commitment (the complement-taking
predicate) and the content of commitment (the complement), the more the
construction encoding them appears as a single clause. In this way, signs of
intentional status can be ordered as a function of the tightness between the
mode of commitment and the content of commitment. This is shown in
figure 2. It thus appears that the best way to group and order the propositional expression of intentional status is by the degree to which the intentional state (i.e., mode of commitment) is implicated in the intentional object (i.e., content of commitment event). Phrased in terms of embodied

Representations of the world

93

Figure 2. Relative tightness of interclausal relations involving Mode of Commitment Events

(E MC ) and Content of Commitment Events (E CC ), where Mode of Commitment is Want,
Remember, and Know, and Content of Commitment is Does It. Constructions Involving
Want are Tighter than Constructions Involving Remember, which are Tighter than Constructions Involving Know.

signs and antecedent and consequent sign events, this is the question of
how iconic-indexical (or iconically overlapping in quality and indexically
related though causality) the embodied sign is with either the antecedent
sign event (I saw him do it) or the consequent sign event (I want to do it).
In short, by attending to the grammar of certain linguistically-encoded
signs of intentional statuses, one has a way of accounting for the genus of
intentionality itself (complement-taking predicates), various species of intentionality (complement-taking predicates grouped as a function of their
relative tightness), and the logic of intentionality (how much causal/ontological overlap there is between a mode of commitment and a content of
commitment). This point has broader implications insofar as understandings of intentionality (as a putative psychological phenomenon) are often
grounded in, if not derived from, these overt linguistic encodings. This
conclusion is surprisingly absent in works by philosophers of Mind, although their data is primarily morphosyntactic (cf. Brentano 1995 [1874];
Ryle 1984 [1949]; Searle 1983). As will be seen in sections 6 and 7, the
lexicalization of modes of commitment is a condition for speakers understanding of intentionality: hence, these facts will have ramifications for
speakers theories of mind insofar as their intuitions about mind are so
often grounded in their experience with language.

6. Beliefs, perceptions, and intentions (knowledge, memories, plans,

wishes)
There are a number of ways to characterize a proposition. Propositions
can serve as the premises and conclusions of inferences. For example, all

94

P. Kockelman

p are q, John is a p, therefore John is a q. This stresses the relation of

propositions to logic. Propositions can serve both as reasons (for other
propositions) and in need of reasons (by other propositions). For example, if someone asserts p, one can ask for a reason why; and if asked a
reason why (someone said or did something), one can assert p. This is
equivalent to the first point, but stresses the relation of propositions to rationality. Propositions can be expressed by assertions. For example, John
is a dentist. This stresses the relation of propositions to language, specifically the seemingly most unmarked form of utterances, the declarative
sentence. Propositions can represent states of aairs. This stresses the relation of propositions to res cognitans, representationalism, and Mind
more generally. Propositions can be ascribed by that-clauses. For example, I believe that John is a dentist. This stresses the relation of propositions to complementation, sense and reference in Freges sense, and Brentanos classic definition of intentionality as object-directedness more
generally. Propositions may be understood as the contents of assertions,
commands, or questions which have been shorn of their illocutionary
force. For example, sentences with declarative, interrogative, and imperative illocutionary force, respectively, can have identical propositional contents: you will go to the store, will you go to the store? and go to the
store! And propositions may be understood as the inferentially articulated
objects of assertions. This stresses the relation of propositions to semiosis:
(natural language) signs, (inferentially articulated) objects, and (ultimate
representational) interpretants.
In The semiotic stance, two semiotic styles were discussed: one that
focuses on sign-object relations to the neglect of interpretants; and another that focuses on sign-object-interpretant relations, with an emphasis
on sign-interpretant relations. It was argued that, though both styles are
compatible, the former is by far the most frequently used (as exemplified
in Saussurian semiology and modern semantics), and tends to objectify
the object and subjectify the interpretant (if taking the latter into account
at all). The approach used here, one that focuses on sign-interpretant relations in the larger context of sign-object-interpretant relations, tends to
objectify the interpretant (making it as object-like as signs) and intersubjectify the object (making it relatively dependent on symmetric attitudes
within a community). When the objects in question have propositional
content, a similar point can be phrased in terms of representationalism
versus inferentialism. For example, philosophers like Searle will argue
that intentional states have representational contents: they represent
states of aairs in the world (satisfactorily or not).17 In contrast, philosophers like Brandom will shift emphasis from the correctness of representation to the appropriateness of inference. Again, while neither one of

Representations of the world

95

these styles is necessarily prior, the former has had explanatory privilege
over the latter in western philosophy and linguistics, while the latter has
been underappreciated and undertheorized (Brandom 1994). In this section, the sign-interpretant approach to semiotics introduced in earlier essays is further articulated in terms of propositionally contentful signs and
interpretants. The key text is Brandoms Making it Explicit, and various
authors whom Brandom is in indebted to or in dialogue with from
Kant, Frege, Wittgenstein, Sellars and Dummett to Anscombe, Davidson, Austin, Grice and Searle. Nonetheless, none of these scholars would
probably agree with precisely this formulation.
In light of these theoretical commitments, there are three basic constituents of the representational whole: observations, assertions, and actions
(see Brandom 1994: chapters 3 and 4). Each of these constituents is a semiotic process, and hence has a sign, an object, and an interpretant. As
with the constituents of the residential whole, the constituents of the representational whole are mainly defined via their objects in this case, intentional statuses. Intentional statuses involve a mode of commitment and
a content of commitment. The content of commitment is just a proposition. And the mode of commitment is a way of relating to the content of
commitment. In this regard, the intentional status, or object-component,
of an assertion is an epistemic commitment (to a propositional content).18
The intentional status of an observation is an empirical commitment (to a
propositional content). And the intentional status of an action is a practical commitment (to a propositional content). When these modes of commitment are lexicalized, they are often called propositional attitudes
in this case, the words believe, perceive, and intend. And when these
modes of commitment are psychologized, they are often called psychological states in this case, belief, perception, and intention. (While
there are other modes of commitment, for present purposes these are the
three most important kinds.)19
If an epistemic, empirical, or practical commitment to a proposition is
an intentional status, it has an intentional role: any enactment of that
commitment to a proposition. To have an epistemic, empirical, or practical commitment to a propositional content, and hence to hold a particular intentional status, is to be committed and entitled to certain modes of
signifying and interpreting in particular, those modes of signifying and
interpreting that logically and causally, or inferentially and indexically,
cohere with ones epistemic, empirical, and practical commitments. To
assert (e.g., utter a declarative sentence), to observe (e.g. attend to a state
of aairs), and to act (e.g., engage in a controlled behavior) are perhaps
the most emblematic roles of these intentional statuses. In particular, to
assert, observe or act is to undertake an epistemic, empirical, or practical

96

P. Kockelman

commitment and thereby license others (by making it appropriate for

others) to attribute such a commitment (and sanction ones behavior accordingly). Other behaviors and utterances may lead others to attribute
an intentional status, as seen in section 3, but they do not necessarily
make it appropriate for others to make that attribution. In short, keep
distinct intentional roles in general (any utterance or behavior that could
lead one to attribute an intentional status to another), and emblematic
intentional roles: those utterings of declarative sentences, attendings to
states of aairs, and undertakings of controlled behaviors that license
others to attribute an intentional status.
Brandom (1994) has introduced four useful terms in this regard: undertaking, attributing, acknowledging, and ascribing. As just mentioned, to
undertake an intentional status (or, more precisely, a mode of commitment to a propositional content), is to engage in some intentional role
that licenses, or entitles others to attribute it to you. For the constituents
of the representational whole, the emblematic intentional roles are uttering a declarative sentence (in the case of epistemic commitments, or belief ), engaging in a controlled behavior (in the case of practical commitments, or intention), and attending to a state of aairs (in the case of
empirical commitments, or perception). To attribute an intentional status is just to take another to have an intentional status, as evinced in ones
sanctioning practices. To acknowledge an intentional status is to attribute
an intentional status to oneself, and hence to engage in self-sanctioning
practices. And to ascribe an intentional status is to explicitly attribute
both a mode of commitment and a content of commitment to another
say, with a propositional attitude: John believes shes a witch and
hence simultaneously attribute an intentional status (to another i.e.,
what John believes) and undertake an intentional status (oneself i.e.,
what one believes about what John believes).20 Ascriptions will be treated
in detail in section 6.
A belief in the psychological sense is best understood as acknowledgement, or self-attribution, of an epistemic commitment: one attributes an
epistemic commitment to oneself, such that one can self-sanction as a
function of whether ones behavior conforms with that epistemic commitment or not.21 An intention in the psychological sense is best understood
as acknowledgement of a practical commitment: one attributes a practical
commitment to oneself, such that one can self-sanction as a function of
whether ones behavior conforms with that practical commitment or not.
And a perception in the psychology sense is best understood as acknowledgment of an empirical commitment: one attributes an empirical commitment to oneself, such that one can self-sanction as a function of
whether ones behavior conforms with that empirical commitment or

Representations of the world

Table 2.

97

Inferential and indexical articulation of intentional statuses

Couched as Semiotic Process

Observation

Assertion

Action

Couched as Mode of
Commitment

Empirical

Epistemic

Practical

Couched as Mental State

Stand as Reason
Stand in Need of Reason
Caused by State of Aairs
Causal of State of Aairs
Non-displaced causality
Displaced causality

Memory
X

Perception
X

Belief
X
X

Intention

Plan

X
X
X

X
X

not. Such modes of self-attribution, or acknowledgment, of epistemic,

practical and empirical commitments pick up the reflexive sense of intentional states that is classically attributed to believing, intending, and
perceiving. Thus, acknowledging an epistemic, practical, or empirical
commitment is inferentialisms equivalent to having a belief, intention,
or perception. In short, while intention, perception, and belief as theorized here look and behave like mental states, they are theorized in a radically dierent idiom.
The three key constituents of the representational whole dier in
regard to the inferential and indexical articulation of their objects.22 Assertions (or the undertaking of epistemic commitments) can stand as reasons, and in need of reasons. Observations (or the undertaking of empirical commitments) can stand as reasons, but not in need of reasons; and
they are indexically caused by a state of aairs. And actions (or the undertaking of practical commitments) can stand in need of reasons, but
not as reasons; and they are indexically causal of a state of aairs.23 See
table 2.
These points should be developed in detail. Insofar as empirical commitments can stand as reasons they have propositional contents; and insofar as they reliably stand in contiguity with the states of aairs represented by their propositional contents, they have empirical contents. For
example, ones perception of a state of aairs must be caused by that state
of aairs. That is, the empirical commitment to the propositional content
must be indexically connected to (in the sense of caused by) the state of
aairs represented by that content; and the empirical commitment to the
propositional content may be inferentially grounding of an assertion (or
epistemic commitment) it can justify a belief.24 Insofar as practical
commitments can stand in need of reasons they have propositional contents; and insofar as they reliably stand in contiguity with the states of

98

P. Kockelman

aairs represented by their propositional contents, they have practical

contents. For example, ones intention to undertake a state of aairs
must cause that state of aairs. That is, the practical commitment to the
propositional content must be indexically connected to (in the sense of
causal of ) the state of aairs represented by that content; and the practical commitment to the propositional content may be inferentially
grounded in an assertion (or epistemic commitment) it can be justified
by a belief. Assertions are inferentially grounding of and grounded in other
assertions. In this way, assertions have both modes of inferential grounding, but neither practical nor empirical contents being neither (directly)
caused by, nor (directly) causal of, states of aairs (but they can both justify a belief and be justified by a belief ).25
As made famous by Searle (1983), and as rearticulated by Brandom,
there are two types of practical commitment, and hence two types of
intention: intentions in action and prior intentions. Intentions in action
are distinguished by the fact that the action itself (or controlled behavior) is the acknowledgment of the practical commitment. For prior intentions (which are often called plans), the acknowledgment of the
practical commitment can come anytime before the action itself (for example, by a promise or an I shall assertion). Both prior intentions and
intentions in action must have practical contents that (reliably) stand in
contiguity with the states of aairs represented by their propositional
contents either immediate contiguity in the case of intentions in
action, or displaced contiguity in the case of prior intentions. It is precisely because of displaced contiguity (or non-immediate indexicality)
that prior intentions (or plans) more easily miss their mark, and are
more often subject to dispute, than intentions in actions. Prior intentions
specify the action one is committing oneself to in general terms; intentions
in action specify the action in demonstrative terms (Brandom 1994: 257).
A prior intention or plan matures into an intention in action (Brandom
1994: 257). Thus, intentions in action are often prior intentions whose
time has come.26
Analogously, it may be argued that there are two types of empirical
commitment, and hence two types of perception: perceptions in observation and subsequent perceptions (cf. Grice 1989; Searle 1983). Perceptions
in observation are distinguished by the fact that the observation itself (or
attention to a state of aairs) is the acknowledgment of the empirical
commitment. For subsequent perceptions (which may be called memories), the acknowledgement of the empirical commitment can come anytime after the observation itself (for example, by a statement like I saw
. . . or I remember . . .). Both subsequent perceptions and perceptions
in observation must have empirical contents that (reliably) stand in

Representations of the world

99

contiguity with the states of aairs represented by their propositional

contents either immediate contiguity in the case of perceptions in observation, or displaced contiguity (or non-immediate indexicality) in the
case of subsequent perceptions. It is precisely because of the displaced
contiguity that subsequent perceptions (or memories) more easily miss
their mark, and are more often subject to dispute, than perceptions in observation. Echoing Brandoms ideas regarding the relation between intentions in action and prior intentions, subsequent perceptions specify
the observation one is committing oneself to in general terms; perceptions
in observation specify it in demonstrative terms. A perception in observation matures into a subsequent observation or memory. A subsequent
perception is a perception in observation whose time has past.
To attribute reliability to an actor or observer is to undertake an epistemic commitment to the propositional contents of their practical or empirical commitments (see Brandom 1994: 206212; and Sellars 1997
[1956]). More narrowly characterized, to attribute reliability to an observer is to attribute a disposition to respond to a state of aairs by bringing about (normatively) an empirical commitment to a proposition representing that state of aairs. And to attribute reliability to an actor is to
attribute a disposition to respond to a practical commitment to a proposition by bringing about (normatively) the state of aairs represented by
that proposition. In the case of intentions in action and perceptions in observation, one can test the reliability of the actor or observer by examining the state of aairs itself (either the one that brought about the observation, or the one that was brought about by the action). On the one
hand, states of aairs normatively bring about epistemic commitments
to propositions that represent those states of aairs; on the other hand,
practical commitments to propositions normatively bring about the states
of aairs represented by those propositions. In this way, reliability licenses
others to undertake epistemic commitments to the empirical and practical
commitments undertaken by observers and actors, respectively. Thus,
ones acknowledging an empirical commitment to the eect that there is
a red thing in front of one is good reason for another to undertake an
epistemic commitment to the eect that there is a red thing in front of
one (Brandom 1994: 236). And ones acknowledgment of a practical
commitment to raise ones arm is good reason for another to undertake
an epistemic commitment to the eect that one will raise ones arm
(Brandom 1994: 236). Notice, then, how reliability licenses inferences
inter-personally, not just intra-personally. Needless to say, memories (or
subsequent perceptions) and plans (or prior intentions) are much less reliable than perceptions in observation and intentions in action precisely
because of their displaced contiguity.

100

P. Kockelman

So far the focus has been on intra-personal inferential articulation: how

being entitled or committed to some propositional content entitles or
commits one to other propositional contents. With assertions, inferential
articulation is both intra-personal and inter-personal. As Brandom sees it,
the inter-personal inferential articulation of propositional contents is the
way in which being committed or entitled to a proposition may commit
or entitle others to a proposition (1994: 168175). For example, if one
makes an assertion, and thereby undertakes an epistemic commitment to
the propositional content of that assertion, not only does one commit to
the propositions that logically follow from that commitment (via intrapersonal inferential articulation), and not only does one entitle others to
attribute that commitment (and its logical consequences) to one, but one
also entitles others to undertake that commitment (and its logical consequences) themselves, and one entitles others to inquire into ones entitlement to that commitment. This entitling of others to undertake ones
commitment may be called discursive authority. And this entitling of
others to inquire into ones entitlement to that commitment may be called
discursive responsibility. Thus, if discursive authority involves having
ones commitment provide reasons for others commitments, discursive
responsibility involves having to provide reasons for ones commitments
to others. In this framework, assertions are crucial then because of the
structure of their intra- and inter-personal inferential articulation: another
can use ones assertion as a reason (authority); or another can demand a
reason for ones assertion (responsibility).
Knowledge is usually taken to mean justified true belief (Brandom
1994: 202): that is, the attribution of an epistemic commitment (i.e., belief ); the attribution of being entitled to that commitment (i.e., justified); and the acknowledgment by the attributor of that commitment
(i.e., true). Assertions, being inter-personally and intra-personally inferentially articulated, are fundamentally knowledge claims. And believing
is dependent on knowing as a sort of attribution of commitment without
attribution of entitlement to commitment and without acknowledgment
of commitment (oneself ). Insofar as epistemic commitments are dependent on assertions, and insofar as assertions are implicated in knowledge
(via inter- and intra-personal inferential articulation), this says that knowing is prior to believing. Indeed, belief is a kind of knowledge without discursive responsibility (to be entitled) or discursive authority (to entitle
others). Somewhat paradoxically (if one ignores the inter-personal inheritance of inferential articulation), it is not that knowing is an upgraded
form of believing, it is that believing is a downgraded form of knowing.
Not only is knowledge prior to belief, but belief is prior to perception
and intention or, rather, epistemic commitments are prior to empirical

Representations of the world

101

and practical commitments. This should come as no surprise insofar as

declarative sentences are the only constituents of the representational
whole whose propositional contents are explicit. (Though in ascriptions,
one may explicitly assert what the observer is perceiving, or what the
actor is intending). This is, of course, why Anscombes intentions are actions under a description: intentions require representational interpretants. And as intentions are actions under a description (or within a representational interpretant, as argued at the end of Residence in the
world), perceptions may be understood as observations under a description (or within a representational interpretant). That is, the movement
from mere purchase and purpose to perception and intention or from
sentience to sapience requires assertions. This is just to say that assertions (qua representational interpretants) confer propositional content on
purchases and purposes. Of course, purposes (and purchases) are prior to
assertions. Notice, then, that from the standpoint of this theory, knowledge
is prior to (or more originary than) belief, belief is prior to perception and
intention, and purchases and purposes are prior to belief.
Justifying a practical commitment consists in exhibiting a suitable piece
of practical reasoning in which it figures as a conclusion (Brandom 1994:
244).27 This follows a Kantian lead in which to treat a performance as
an action is to treat it as something for which it is appropriate to demand
a reason. If undertaking an epistemic commitment entitled others to attribute it to you, and thereby involved a conditional responsibility to demonstrate entitlement (e.g., provide a reason for ones assertion), then to
undertake a practical commitment involves a conditional responsibility
to demonstrate entitlement: in particular, to exhibit a suitable piece of
practical reasoning for which it figures as a conclusion (Brandom 1994:
245). In particular, an intention should be understood as the conclusion
of an inference that has two parts: an epistemic commitment and a proattitude, where the latter can be a personal preference (particular to an
individual), a status commitment (particular to a role), or a value commitment (particular to an identity). For example, an intentional action
of opening an umbrella can be rationalized by attributing to the actor an
epistemic commitment that it is raining, and either a personal preference
or wish (say, to stay dry), a role commitment or status (say, one must
keep ones uniform clean), or an identity commitment or value (say, dryness is godliness). In this way, intentions intrinsically relate to wishes,
(social) statuses, and values (as species of pro-attitudes which might explain them); and intentions intrinsically relate to reason-giving, and rationality more generally. (Though note that these pro-attitudes are usually
implicit, and only become explicit in cases of strange actions: why did
she do that?) Here, then, is another key locale where identities and roles,

102

P. Kockelman

as theorized in Residence in the world, are expressed in modes of

representation.
Regarding personal preferences or wishes, what is crucial is not where
they reside nor what they are, but that they are person-specific rather than
role-specific or identity-specific. (They nevertheless can be people-specific:
say, to seek pleasure and avoid pain.) These wishes will be left relatively
untheorized here. Nonetheless, note the following characteristics of them:
one, as just discussed, they should be defined in relation to pro-attitudes
and primary reasons (and they should be distinguished from social statuses and values); two, their emblematic roles are optative utterances
(e.g., if only it would rain), and they are attributed to others with desire
and wish predicates (e.g., she wants to be a witch, or he wishes he were a
warlock); three, in contrast to perceptions and intentions, they seem to
lack indexical articulation; four, no distinction is being made between desiring and wishing; and five, they relate to selfhood (as turning on what
would preserve the boundaries of the self: e.g., in general one wishes for,
or desires, that which allows the self to wax and/or that which prevents
the self from waning).28
These three pro-attitudes (wishes, statuses, and values) are all species of
a common genus: they are to be defined by the way they complete primary reasons.29 In particular, pro-attitudes link epistemic commitments
and practical commitments (or beliefs and intentions). For Brandom,
a pro-attitude is what turns a materially good inference into a formally
good inference: it makes explicit (by expressing in the form of a claim)
what is implicit (in the endorsement of a claim). Indeed, it makes explicit
the endorsement of a pattern of inferences: if one is entitled to an epistemic commitment, then to inherit entitlement to a practical commitment is
to implicitly attribute a pro-attitude. This gives rise to patterns: for that
pro-attitude will allow for inheritance from entitlement to epistemic commitments to entitlement to practical commitments in many situations.
These patterns are, of course, just the key locale where wishes, statuses,
and values come to the fore: they provide the best evidence for the existence of such semiotic objects especially when the patterns are
violated.30
There is nothing corresponding to discursive authority in the case of
practical commitments (Brandom 1994: 253). Thus, while one can entitle
others to ones epistemic commitments by undertaking them (say, one
makes an assertion), one cannot entitle others to ones practical commitments by undertaking them (say, one engages in an action). In part, this is
a consequence of the fact that intentions are typically intrapersonal,
whereas beliefs are interpersonal. Nonetheless, in the case of individuals
with the same role, and hence the same status, ones status provides a

Representations of the world

103

reason for actions, and hence can count as a reason for the others actions. And, in the case of individuals with the same (group) identity, and
hence the same value(s), insofar as ones identity can provide a reason for
ones actions, it can act as a reason for the others actions. In this way, in
the case of shared roles and shared identities, ones practical commitments
can provide reasons for anothers practical commitments. Indeed, particular identities that purport or attempt to be transcendental identities involve values that all rational people should commit to (e.g., the categorical imperative or the golden rule).31
Finally, just as questions of coherence arose in the residential whole,
they arise in the representational whole. In particular, there are five kinds
of coherence that can be defined in terms of the foregoing discussion:
rational, causal, representational, intersubjective, and sincerity. Rational
coherence (or inferential coherence) is just the degree to which commitments that may stand as reasons or in need of reasons actually do. What
is the reason for ones epistemic commitment, and what is ones epistemic
commitment a reason for? What is the reason for ones practical commitment? And what is ones empirical commitment a reason for? Causal coherence (or indexical coherence) is the degree to which the states of aairs
represented by the propositional content of empirical commitments actually caused the empirical commitment, or the degree to which the states
of aairs represented by the propositional content of practical commitments actually cause the state of aairs. Was a perception caused by a
state of aairs, and was an action causal of a state of aairs? Representational coherence is the degree to which commitments are satisfied.32 This
is not the causality of the connection but the correctness of the connection. Are beliefs true? Are perceptions veridical? Are intentions sated?
Intersubjective coherence is the degree to which intentional statuses attributed are acknowledged, and the degree to which intentional statuses undertaken are attributed. Are ones understanding of ones own intentional
statuses in agreement with others understanding of them? And sincerity
coherence is the degree to which the intentional status undertaken is acknowledged. Does one believe what one asserts, intend what one promises to do, remember what one recounts?
Insofar as the constituents of the representational whole are holistically
governed, these five kinds of coherence can exist (or not) at many dierent levels: representational whole, institutional whole, situational whole,
intersubjective whole, or experiential whole. Finally, by focusing on coherence is not to stress that coherence is unmarked (or the usual everyday
case): the representational whole can be massively incoherent globally
or locally. The emphasis is rather that coherence is a normative question
that exists in (at least) five dimensions. Just as in the case of coherence

104

P. Kockelman

discussed in Residence in the world, to find more coherence than is (socially) warranted is a key feature of fetishization.

7.

Predicating intentional statuses of people and properties of intentional

statuses

As seen in the last section, intentional statuses involve both modes of

commitment (how one relates to a propositional content i.e., whether
one believes, intends, or perceives) and contents of commitment (the
propositional content one relates to i.e., what one believes, intends, or
perceives). What is of interest in this section is how modes of commitment
become contents of commitment. Phrased in a psychological idiom, this
is equivalent to intentional states becoming intentional objects. And
phrased in a semiotic idiom, this is equivalent to having propositional
contents conferred on modes of commitment by representational interpretants of them typically via processes of ascription. Insofar as propositional contents are the inferentially articulated objects of assertions,
this means three things: 1) examining the words that refer to intentional
statuses (e.g., believe, intend, fear); 2) examining the utterances that predicate intentional statuses of people (e.g., John is angry); and 3) examining
the utterances that predicate properties of intentional statuses (e.g., anger
is an emotion). Such referring and predicating expressions or intentional status predicates (ISPs) and the utterances in which they are
implicated introduce a new order of mediation into intentionality.
While this order of mediation is introduced with any representational interpretant (compare the conferral of propositional contents on constituents of the residential whole via words like tree, hammer, run, mother,
and Christian), it is worth drawing out the repercussions for intentionality. There are two key processes, lexicalization and displacement, each of
which involves a number of interrelated sub-processes.
Lexicalization is the process whereby the signs of modes of commitment become lexical predicates (rather than grammatical operators).
Such lexemes were referred to as ISPs. For example, rather than undertaking an empirical commitment by attending to a state of aairs (say,
physically looking in a certain direction), one undertakes an empirical
commitment by asserting I see that . . . or I remember that . . . Or rather
than undertaking a practical commitment by engaging in a controlled behavior, one undertakes a practical commitment by asserting I intend to
. . . or I plan to . . . An analogy may be made with speech acts. In particular, most utterances have illocutionary force and propositional content: one asks a question, issues a command, or makes an assertion.

Representations of the world

105

What is crucial about speech act predicates (e.g., say, ask, order, and
so forth), is that they confer propositional content on illocutionary force.
And just as there are usually only a handful of grammatical operators for
indexing illocutionary force (optatives, exclamatives, imperatives, interrogatives, and declaratives), there may be hundreds of lexical predicates
for referring to it. The movement from illocutionary force (or grammatical operator) to explicit performative (or lexical predicate) is a movement
from closed class to open class. In the case of intentional statuses, as in
the case of speech acts, this process involves a number of important and
interrelated features: conceptualization, semantic fields, language-internal
glossing, metaphorical elaboration, and projection.
First, conceptualization is the process whereby modes of commitment
come to have conceptual content, insofar as the ISPs that refer to them
are implicated (as referring and predicating expressions) in utterances
that have propositional content. Insofar as concepts expressed by ISPs
are the inferentially articulated objects of ISPs, one may inquire into the
conceptual structure of ISPs like belief, intention, and fear. In particular, a number of questions can be posed. Are the conceptual contents of
certain ISPs more basic or primary than the conceptual content of other
ISPs? For example, does knowing necessary depend on believing, but
not vice versa; or does lust necessarily depend on desire, but not vice
versa? Are there basic ISPs ones whose conceptual structures does not
depend on the conceptual structure of other ISPs, but whose conceptual
structure other ISPs depend on? How does the conceptual structure of
such predicates dier as a function of whether the speaker using them is
an expert or not? How should these words be translated or calibrated
across languages, cultures, and eras? Are some concepts found in all languages, cultures, and eras? One can ask about how these concepts are
structured. For example, are emotion concepts structured like a script?
Are there basic level terms? Are expert definitions structured like Aristotelian categories? Finally, one may ask how the conceptual structure of
ISPs relate to the practical content of intentional statuses. For example,
if emotions are structured like a script, how does this script-like structure
relate to roots and fruits of emotional statuses meaning the actual sign
events that lead to them and follow from them?
Second, in the movement from grammatical operator to lexical predicate, there may be a large number of ISPs, constituting a kind of semantic
field of modes of commitment (i.e., a lexicon of mental states), which are
all related to each other via relations like synonymy, antonymy, partonomy, taxonomy, and so forth. In this way, a menagerie of putative mental
kinds acquires a kind of objectivity so far as these ISPs seem both to refer
to concrete things, and to belong to a common genus. For example, just

106

P. Kockelman

as there are dierent kinds of fruit (bananas, apples, oranges) and different kinds of food more generally (fruit, vegetables, etc.), there are different kinds of emotions (say, fear, joy, anger, etc.) and dierent kinds of
intentional statuses more generally (say, the volitive, the cognitive, and
the emotive).
Third, so far as these ISPs have conceptual content, this semantic field
of intentional statuses allows for language-internal glossing. For example,
belief is a kind of knowledge, there are two key components to mind:
emotion and cognition, fear and shame are types of emotion, desire
and belief are basic intentional states, and so on. Peoples understanding
of intentional statuses, or mental states, can be articulated and explicit
rather than just intuitive and implicit. There are dictionaries, and selfhelp books, basic psychology texts, and maxims. Again, a kind of objectivity is introduced so far as speakers can define, and hence regiment and
standardize, the definitions of intentional statuses.
Fourth, these ISPs are often derived through metaphorical elaboration.
For example, among the Qeqchi-Maya (Kockelman 2003b, 2005b),
most ISPs are articulated in terms of relatively concrete processes involving the heart: for the heart to shrink is to become afraid; for the heart to
double is to become conflicted; for the heart to be red is to be jealous; for
something to get lost in the heart is to forget it; and so on. In this way,
lexical constructions expressing relatively abstract concepts (intentional
statuses) are metaphorically articulated in terms of lexical constructions
expressing relatively concrete concepts (color, size, shape, number, etc.).
In this way, relatively novel and abstract concepts can be readily introduced, and readily understood, in terms of the properties of the relatively
old and concrete domain the terms were borrowed from. And, in this
way, inferences appropriate in the concrete domain may be extended to
be appropriate in the abstract domain. This metaphorical construal of
the abstract through the concrete is another process whereby intentional
statuses acquire a kind of objectivity.
And fifth, lexicalization may lead to projection: the process whereby
features belonging to the signs (of objects) are projected onto the objects
(of signs). In regard to the representational whole, projection is the way in
which linguistic or semiotic features of ISPs are understood as ontological
or natural features of the intentional statuses that such ISPs seem to refer
to (cf. Whorf 1956 [1939]). In certain cases, such as explicit metaphor, this
process is relatively trivial: for example, the degree to which one actually
takes ones heart to be red when one is jealous, or to double when one is
conflicted. In other cases, involving grammatical features, this is less trivial: for example, the degree to which one takes intentional statuses to be
ontological states (believe), state-changes (become angry), or activities

Representations of the world

107

(think) insofar as the ISPs that refer to these statuses are semantic states,
state-changes, or activities. Indeed, as discussed in section 4, intentionality as a genus-level phenomenon is often phrased (following Brentano) as
object-directedness, and exemplified by listing the ISPs that are all just
complement-taking predicates. That is, the object-directedness of mental
states is related to the complement-takingness of intentional status predicates. And notice how many philosophers will take the intentionality of
mental states to be equal to the intentionality of speech acts, and theorize all the properties of mental states using evidence derived from speech
acts and then take these features to be fundamental to mental states
rather than to speech acts (see Searle 1983 as the canonical example). In
any case, this is yet another process whereby the putative referents of ISPs
acquire a kind of objectivity through the phenomenal, structural, and semantic properties of the signs themselves.
In conjunction with lexicalization (and its attendant features: conceptualization, semantic fields, language-internal glossing, metaphor, and projection) is displacement: the way that ISPs, insofar as they are lexical
predicates (such as nouns and verbs), may take grammatical operators
that displace them in space, time, possibility, and person. That is, ISPs
such as believe, fear, and desire can be attributed to oneself and
others, in the present, past or future, and in actual, possible or counterfactual worlds. For example, one can say I believe that . . . , you believe
that . . . , and she believes that . . . One can say I fear that . . . , I feared
that . . . , and I will fear that . . . And one can say I want to . . . , I may
want to . . . , and I would have wanted to . . . Displacement is a function of the fact that ISPs are operated on by grammatical categories
that are shifters person, tense, mood, and status. These shifters calibrate the spatial-temporal-logical-personal position of the narrated event
relative to the speech event (see Goman 1981 [1979]; Jakobson 1990).
In this way, ISPs allow one to displace the mode of commitment event
from the sign event, or the event of having an intentional status from the
event of indicating that one has it. This giving of the mode of commitment an event-like character (being positioned in space and time, and
being particular to a specific person or a possible world) is another way
in which intentional statuses acquire a kind of objectivity. Displacement is related to several other properties which should be discussed
in detail: tightness, ascription, transparency, meta-intentionality, and
assertability.
First, if one adds to the discussion of displacement the account of tightness oered in section 4, one sees that there are really three events that
may be positioned relative to each other: the sign event (or the event
in which one expresses an utterance that involves an ISP); the mode of

108

P. Kockelman

commitment event (or the event in which one is committed to some

propositional content); and the content of commitment event (or the
event one is committed to). For example, in the utterance you believed
she went to the store, her going to the store is the content of commitment
event, your believing this is the mode of commitment event, and the uttering of this assertion is the sign event. All of these events can involve displacement in person, space, time, and possibility. Section 4 described the
ways in which mode of commitment events relate to content of commitment events via tightness: the degree to which the mode of commitment
event and the content of commitment event are ontologically a single
event. In contrast, displacement is the degree to which the mode of commitment event and the sign event are ontologically a single event. If tightness is a relation between the mode of commitment event and the content
of commitment event, displacement is a relation between the sign event
and the mode of commitment event (and often the content of commitment event).
Second, ascription is the process whereby an ISP is explicitly predicated
of another person (or oneself ), thereby providing a representational interpretant of anothers (or ones own) intentional status. As Brandom sees it,
Ascriptions are propositionally explicit attributions (1994: 504). And
Ascribing is attributing one commitment (to another), while undertaking
(acknowledging) a dierent commitment (oneself ) (1994: 504). It is the
species of displacement that turns on the grammatical category of person
(rather than other shifters like deictics, tense, or status). What is crucial
about ascription is that in attributing an intentional status to another
one is undertaking an intentional status oneself. In particular, in the case
of the intentional status ascribed, both the mode of commitment and the
content of commitment are explicit; in the case of the intentional status
undertaken, only the content of commitment is explicit (which is equal
to the mode and content of the commitment attributed). For example, in
saying John hopes that Woody Allen will run for governor, one attributes to John a mode of commitment (hope) and a content of commitment
(that Woody Allen will run for governor); and one undertakes a mode of
commitment (an epistemic commitment or belief ) to another content of
commitment (that John hopes that Woody Allen will run for governor).
Ascription is crucial insofar as it is the site in which dierences between
ways of representing the world most forcefully arise between what is
attributed to another and what is undertaken oneself. Hence, debates
can arise as to what is the correct or true representation. Indeed, our
very sense of the subjectivity of mental states is just that dierent people
can be committed to contradictory and/or incoherent propositional contents. There is evidence that such discrepancies (between commitments

Representations of the world

109

undertaken and attributed) arise in the context of arguing over states of

aairs in the world, as with sibling rivalry (see Tomasello 1999, and references therein); and one suspects that the ability to track such discrepancies is crucial for passing theory of mind tests.33
Third, transparency characterizes the degree to which the semiotic
means of undertaking (or expressing) an intentional status is the same as,
or similar to, the semiotic means of ascribing (or describing) an intentional status. Silverstein (1981) introduced this idea (which he called
metapragmatic transparency) to describe how explicit performatives
(e.g., words like baptize, wed, bet, promise, etc.) can be used both to perform speech acts (e.g., I promise that . . .) and to describe speech acts
performed (e.g., he promised that . . .). And he argued that the more
transparent a sign is, the more easily speakers can provide an interpretation of its meaning in the sense of what must be the case for it to be
used appropriately, and what comes to be the case if it is used eectively.
The same distinction is partially operative here: ISPs can be used both to
undertake (or express) an intentional status and ascribe (or describe) an
intentional status. For example, the ISP believe is transparent: it can be
used both to undertake an intentional status (I believe that . . .) and to
ascribe an intentional status (he believed that . . .). Contrast this with interjections, or various other non-propositional signs, that can be used to
undertake an intentional status (e.g., ouch!), but not to ascribe an intentional status (e.g., *he ouched).34 It should be stressed that ISPs are different from explicit performatives in that explicit performatives perform
what they describe by describing it: I can only wed you by saying I wed
you (see Austin 2003 [1955]). Such a reflexive-feature is not usually operative with ISPs.
Fourth, meta-intentionality is the process whereby one has modes of
commitment whose contents of commitment represent other modes of
commitment. For example, one can say I believe she wants a hamburger, or I wish I didnt fear snakes, or Im afraid her beliefs dont
mesh with my own, and so on. Such reflexivity can, of course, go on indefinitely: I believe that he believes that she believes . . . This property
turns on the fact that ISPs typically take complements, and these complements can themselves consist of ISPs, and so on indefinitely. Various
forms of meta-intentionality are often understood to be the sine qua non
of human-specific cognitive processes: choice (desiring particular desires);
empathy (feeling others feelings); conscience (evaluating ones own motivations), and self-knowledge (in the Socratic sense). And these are crucially implicated in various forms of self-reflexivity.
Fifth, assertability (or explicitness) is the process whereby predicating
intentional statuses of people, or predicating properties of intentional

110

P. Kockelman

statuses, involves ISPs that are parts of assertions and hence can both
stand as reasons (for other assertions) and stand in need of reasons (by
other assertions).35 For example, one can ask someone to explain why
they ascribe an intentional status to another person; and one can ascribe
an intentional status to another person to explain why (say, they engaged
in some controlled behavior). In particular, as with any assertion, the
speaker is responsible to justify her entitlement to the epistemic commitment undertaken by that assertion; and the speaker authorizes others to
undertake that epistemic commitment. In this way, questions of discursive responsibility and authority arise as to intentional statuses predicated
of people and properties predicated of intentional statuses. For example,
arguments may arise as to what counts as good evidence for ascribing an
intentional status, or what behavior may be predicted if someone has
been ascribed an intentional status, and hence one may debate what are
rational and irrational intentional statuses given a persons experience.
Such assertions are therefore subject to challenge, justification, argument,
revision, testimony, repetition, inculcation, extirpation, gossip, and so
forth. In this way, assertions ascribing intentional statuses, and the reasons that lead to them and follow from them, can become fodder for public discussions and political interventions. In the next section, these points
will be generalized.

8.

Empirical investigations, theoretical representations, practical

interventions

In the course of acquiring propositional content through words (ISPs)

that refer to them, intentional statuses may become the object of empirical investigations, theoretical (or epistemic) representations and practical
interventions. This involves two interrelated processes. First, as seen in
the previous section, in addition to predicating intentional statuses of
people via ISPs (e.g., hes angry or he believes that she wants to do
him in), intentional statuses have properties predicated of them via ISPs
(e.g., anger is a negative emotion or belief is a kind of weak knowledge). And second, via both these kinds of predication, modes of commitment (as contents of commitment) are themselves inferentially and indexically articulated. Thus, the following kinds of questions are at issue:
what observations or assertions would entitle or commit one to predicate
property X of intentional status Y, or predicate intentional status W of
person Z; and if one predicates property X of intentional status Y, or
predicates intentional status W of person Z, what assertions or actions
does this entitle or commit one to?

Representations of the world

111

In short, theoretical representations of intentional statuses might be

best understood as assertions (epistemic commitments to propositional
contents) which either represent people as having certain intentional statuses, or represent intentional statuses as having certain properties. Such
theoretical representations can stand as reasons and in need of reasons.
Empirical investigations of intentional statuses might be best understood
as observations (empirical commitments to propositional contents) of the
intentional statuses of people or the properties of intentional statuses.
These can stand as reasons for theoretical representations; and these are
indexically caused by states of aairs. And practical interventions of
intentional statuses might best be understood as actions (practical commitments to propositional contents) which are directed towards aecting the intentional statuses of people or the properties of intentional
statuses. These can stand in need of reasons; and these are indexically
causal of states of aairs. In short, by taking the analysis put forth in
section 5, and reflexively turning it back on itself at the level of mode
of commitment rather than content of commitment, this is a way of
moving from a theory of mind to a theory of truth of a theory of mind,
or from an account of intentionality to an account of the epistemology
of intentionality.
More generally, the term episteme might be used to describe any institution that involves a relatively coherent cluster of such propositional modes of semiosis (e.g., empirical observations, theoretical representations, and practical interventions) in conjunction with a relatively
coherent cluster of non-propositional modes of semiosis (e.g., modes of
comportment turning on aordances, instruments, actions, roles, and
identities).
And, for any episteme, one might ask a number of questions: What
is the topic of the episteme? What questions are members of this episteme trying to answer regarding this topic? What are the stakes of posing such questions (and potentially answering them) for members of
that episteme? How are these topics and questions (and stakes) theoretically elaborated ? That is, how are they conceptually articulated relative to other topics and questions within a research tradition. How are
such theoretically elaborated topics and questions methodologically
grounded ? That is, how are they rendered empirically tractable, in
terms of their conceivable practical bearings. What methods, or instruments and techniques, are used to actually create, collect, store, organize, analyze, interpret, compare, exemplify, present, and communicate
this data? What epistemological practices are implicit in this episteme?
In particular, if knowledge is justified true beliefs, how are beliefs
(i.e., representations of how things stand in some context) turned into:

112

P. Kockelman

1) justified beliefs (i.e., representations that one can be responsible for

asserting if called upon to justify them by some community of scholars);
and 2) true beliefs (i.e., representations that other members of this community will take for granted in their own work, or qualify using their
own research). What ethical practices are implicit in this episteme? If
epistemology is about the responsibility and authority members of an
episteme have (regarding their representations) with respect to other
members of the episteme, ethics is about the authority and responsibility
members have toward those whom it represents, and those for whom it
represents.36
While the term episteme is Foucaults, and while the spirit of this point
is Foucaultian,37 an episteme as used here is grounded in the foregoing
accounts of residence in the world and representations of the world; and
the episteme of interest here is the one whose topics (qua objects of investigation) are intentional statuses per se (rather than, say, electrons,
trees, benzene rings, the common cold, and so forth). In this regard, these
questions might be posed of Freudian psychoanalysis or ethnopsychologies of far-flung peoples, of modern biomedical forms of pill-pushing to
parental wisdom concerning how to soothe the feelings of a distraught
child, of Skinners version of behavioralism to Chomskys version of cognition, from Bettelheims theory of autism to Kohuts theory of the self,
and so on, perhaps indefinitely. While this essay is too circumscribed to
take up any of these questions, there are many monographs that may be
partially understood in this light: from Goman on asylums to Hacking
on multiple personality disorder and fugue, from Foucault on madness
and civilization to Danziger on the historical origins of mainstream psychological research.
For present purposes, one pervasive trend across many of these literatures may be foregrounded: psychologization, or the modern, western understanding of intentional statuses in terms of mental states, and the
range of ontological commitments that go with it. In particular, intentional statuses are: localized (taken to be located in a particular part of a
person say, the mind); privatized (taken to be hidden or invisible in a
way that other phenomena are not); interiorized (taken to be internal
rather than external to a person); subjectified (taken to be more easily in
error, and/or more person-specific, than objective phenomena); dichotomized (understood in terms of a set of oppositions: subjective versus objective, interior versus exterior, private versus public, etc.); individualized
(taken to be held by individuals, rather than larger-than-individual or
less-than-individual entities); moralized (caught up in legal and religious
judgments regarding whether one should be responsible for a behavior
or not, and whether such behavior is good or bad); universalized

Representations of the world

113

(treated as something all human groups have); humanized (treated as

something only human groups have); and humunculized (treated as having agency, or human-like drives, themselves).38 The psychologization of
intentional statuses, then, tends to see them as little pieces of furniture
in the house of Mind. Contrast the categories used to describe intentional statuses that have been put forth in this essay, which treats mental
states in nearly the same terms as it treats social statuses and linguistic
utterances.
Indeed, from a cross-cultural perspective, this psychologization need
not occur at all, or at least not in this particular way. For example, the
ethnographic record (Kockelman 2002) shows that intentional statuses
can be theoretically represented in any number of ways. In particular, intentional statuses in general, or certain kinds of intentional statuses (e.g.,
second-order ones, such as beliefs about beliefs) need not be attributed to
humans, to all humans, and only to humans. Intentionality, or Mind,
need not be understood as a faculty; it may also be understood as, say, a
gift or a curse. And the faculty need not be psychological; it can be physical, theological, or medical.
Intentional statuses need not be attributed to individuals, but can be
shared across individuals, or occur within individuals. For example, there
may be super-individual who hold a single intentional status: couples,
friends, fraternities, nations, tribes, etc. And there may be supraindividuals who hold a single intentional status: faculties (unconscious
drives versus conscious ones), split-personalities, etc. And, indeed, intentional statuses need not be evinced by the same individual who holds
them: ones role may be the primary evidence for anothers intentional
status.
There need not be the same species of intentionality (say, perception,
wish, intention, feeling, and belief ); nor need there be the genus itself (intentionality, qua directedness). Intentionality need not be understood as a
unified domain say the mind module; but may be understood as the
precipitate or eect of many interacting domains. And intentionality
need not be understood as having more basic statuses (say, belief and intention) out of which all others are built, or definable.
Intentional statuses need not be understood as one part of a dichotomy
(mind versus body, or mental versus physical). Intentional statuses need
not be located in some single, internal, particular place (e.g., the mind
or the heart); they can be located outside of the body, in other parts of
the body, or not be localizable at all. Intentional statuses need not be understood as private (relative to sources of evidence for them); they can be
as public as any other phenomenon, or at least no less public than, say,
kinship statuses.

114

P. Kockelman

Intentional statuses may not be caught up in the causal web of ways of

explaining behavior: to explain anothers behavior one makes reference to
their social statuses rather than their intentional statuses i.e., nobody
speculates on others motives. Indeed, there may be injunctions against
ascribing intentional statuses to others.
Intentional statuses need not be linked to morality in the same way
as caught up in responsibility and rights, as judged as good or bad, as
seen as rational or irrational, or as understood as a site where value or
knowledge (about the individual or the world in general) is revealed.
And intentional statuses need not be related to selfhood and agency in
the same way.
Of course, intentional statuses might be theoretically represented in
these ways not because they really are mental states, but because
they have certain semiotic properties. They are embodied signs (and
hence invisible or private) but no dierent from social statuses like
being a mother. And they are propositional (and hence representational)
but no dierent from linguistic expressions. And hence one would do
well to understand mental states as complicated semiotic processes that
stand at the intersection of social statuses and speech acts.

9.

Conclusion: Emotions as natural constructions and social kinds

In The semiotic stance, the following example was given to show the
ways in which any interpretative event may give rise to a number of simultaneous signs: upon hearing a gunshot (as a sign), one may be suffused with adrenaline (aective interpretant); one might make a frightened facial expression (relatively non-purposeful energetic interpretant);
one may run over to look what happened (relatively purposeful energetic
interpretant); one might say that scared the hell out of me (representational interpretant); one may never go into that part of the woods again
(ultimate interpretant); and one might forever believe that the woods are
filled with dangerous men (ultimate representational interpretant). This
example was provided to show that most interpretants are not mental
signs, or even subsequent utterances (a` la answers to questions), but various modes of embodied comportment: feelings, (re)actions, assertions,
and habits. And it was noted that all these interpretants are, through
semiotic framing, just signs (or dispositions to signify) that themselves
can be interpreted by others indeed, they are often bundled together
as evidence for a single ascription: Jake must be terrified of the woods.
To this set of interpretants, one may now add the local episteme in
which terror is understood (say, as a psychological entity that is relatively

Representations of the world

115

uncontrolled, subjective, and natural i.e., an emotion); one might add

Jakes own, simultaneous interpretants of any of these interpretants (as
signs) as what is often understood as the aective experience, or
subjective feeling, that occurs with these other components (e.g., the material qualities of the aective interpretant experienced in light of the
eliciting sign); and finally one might also include within the ultimate interpretant a disposition to reinterpret all signs in a new light (e.g., seeing forests as hiding places for thieves, being spooked by bird calls, trembling at
the mention of the anything arboreal, and so forth).39
As this example demonstrates, The semiotic stance provides a natural
entry into what are usually called the emotions.40 Indeed, emotions as
they have been traditionally theorized are just the relatively systematic
bundling of some of these components (all of which are signs or interpretants of a particular sort: aective, energetic, representational, epistemic, ultimate, and so on). It should be stressed that no single one of these
components is an emotion; rather, any emotion involves all of them.
Moreover, despite the common assumption that emotion is a subjective
state or psychological kind, the ethnographic record shows that secondorder interpretations of this bundling (theorized as epistemes in the last
section) are just as often rendered in moral, spiritual, and physical idioms
as in psychological ones (see Levy 1973; Shweder 1994). From this standpoint, any account of emotions is just an account of how these components articulate with each other (as an ensemble of semiotic processes), in
any sign-community, across sign-communities, and indeed across many
species.
With these general ideas in mind, several caveats should be mentioned.
First, in a monograph entitled What Emotions Really Are, Griths (1997)
argues that the phenomena typically grouped under the term emotion
actually fracture into three ontologically distinct parts, such that the concept itself does not delimit a natural kind.41 He thinks that such distinct
pieces have been grouped together in the past only because they share a
general feature of passivity in contrast to other cognitive phenomena
(in particular, mental states underlying means-end reasoning: e.g., beliefs
and intentions). And, in place of a single category, he argues that one
must (minimally) keep separate aect programs, irruptive motivational
states, and socially-sustained pretense. These should be discussed in turn.
As introduced by Darwin (1965 [1872]), and most elaborately investigated by Ekman (1982), aect programs are short-term, stereotypical responses that involve facial expressions (e.g., non-purposeful energetic
interpretants) and autonomic nervous system arousal (e.g., aective interpretants). Although there is some disagreement on the number of such
programs, most researchers agree on a basic, seemingly cross-cultural

116

P. Kockelman

(and, to some extent, cross-species) set, including anger, fear, disgust, sadness, joy, and surprise.42 Next, irruptive motivational states aect cognitive processes involved in long-term, planned action, and seem to be designed to enforce commitments to strategies that would otherwise be
disrupted by individual calculations of self-interest (see Frank 1988).
These are the most poorly understood; but are essentially complicated ultimate interpretants. Examples of such states are jealousy, guilt, envy,
and shame. Lastly, Griths uses Averills definition of socially-sustained
pretense as a transitory social role . . . that includes an individuals appraisal of the situation, and is interpreted as a passion rather than as an
action (Averill 1980: 312). Thus the aect program of anger is to be
contrasted with the socially-sustained pretense sense of being angry,
which involves a characteristic, but culturally-dependent pattern of behavior that is appropriate in certain situations, and which is thought to
be impossible to control. Socially sustained pretense, then, is just an ultimate (representational) interpretant under another name; and which
comes along with a representational interpretant of the role as relatively
uncontrolled. (And, in general, socially sustained pretence is a covert social construction, in the terms discussed in The semiotic stance.)
Note, then, that although one may be angry as an aect program or
as a socially-sustained pretense or even, perhaps, as an irruptive motivational state, the pathologies of such modalities of anger are distinct (even
if such modalities are referred to with the same term, even if similar conditions may elicit both modalities, even if one modality may influence another, and even if they are culturally valued in similar ways). In this way,
for the purposes of expert reasoning (generalization, induction, etc.),43 it
does not help scholars to group such forms of anger together what one
discovers about one form cannot be used to understand the others (Griffiths 1997). It is for this reason that Griths argues that the category delimited by our everyday concept of emotion does not constitute a natural kind.
Second, assuming that what Griths says is plausible, a key question is
why second-order interpretations, or epistemes, of such disparate domains
take them to belong to a single, unified domain. While most scholars tend
to think emotions have the quality of passivity in contrast to other cognitive phenomena, in a semiotic idiom, one would rather say that semiotic
processes understood in terms of emotion have signs that are relatively
dicult to control, sign-object relations that are relatively dicult to
compose, and interpretants of sign-object relations that are relatively difficult to commit to. Relatedly, insofar as aect programs involve the most
emblematic roles of emotional statuses (that is, facial expressions), and
insofar as aect programs are on the boundary of what is regimented by

Representations of the world

117

natural causes versus cultural norms (and hence what is maximally motivated), properties (and theories) of aect programs are easily projected
onto other seemingly emotional phenomena: naturalness, motivation, uncontrollability, and so forth. And relatedly, emotions seem causally articulated, but not logically articulated caught up in indexicality, but
not inference. In this way, one does not demand a reason for them, or invoke them if a reason is demanded. That is, relatively speaking, they fall
out of the inferential articulation, and inter- and intra-personal inheritance (of commitments and entitlements), that was seen to be fundamental to beliefs, perceptions, and intentions.
And finally, the distinction between natural kinds and social constructions is far too simple. As mentioned in The semiotic stance, a distinction between natural constructions and social kinds should be introduced. In particular, a social kind takes into account the relation between
two relations: or how one relatively motivated set of non-propositional
semiotic processes (say, facial expressions as non-purposeful energetic
interpretants) is articulated in terms of one relatively motivated set
of propositional semiotic processes (say, in the spirit of section 4,
complement-taking predicates that denote emotions as representational
interpretants). Such an abstract formulation of a social kind is a way of
grounding social kinds in reflexive semiosis, or the ways in which signers
use one set of signs to interpret and/or regiment the meaning of another
set of signs in this case, the way both lay-speakers and psychologists
gloss the meaning of facial expressions in terms of a vocabulary of emotion words. One might then use the term natural construction to describe
cross-cultural social kinds: or shared patterns of regimentation linking
sets of propositional and non-propositional semiotic processes, or modes
of residence in the world and modes of representation of the world.
If human beings are those entities whose agency is both enabled and
constrained by the fact that their modes of residence in the world are
never commensurate with their modes of representations of the world),
then the relevant locus for cross-cultural comparison should not be some
mode of residence or representation (e.g., comparing their facial expressions, or comparing their emotion vocabulary) but rather a relationship between the two (e.g., comparing one groups interpretation of facial
expressions in terms of their emotional vocabulary with another groups
interpretation of facial expressions in terms of their emotional vocabulary). So far as human-being is constituted by reflexivity, shared modes
of human-being should turn on shared patterns of reflexivity. Natural
constructions and social kinds, then, are a way of formulating crosscultural and culture-specific processes in terms of relations between residence and representation.

118

P. Kockelman

Notes
1. In certain cases, it may just be a concept rather than a proposition, and hence refer to a
referent rather than represent a state of aairs. For example, I hate cheese or I want my
mother.
2. See Haugeland (1998) for a wonderful summary of various understandings of
intentionality.
3. So, for example, Searle will argue that the intentionality of speech acts is derivative of
the intentionality of mental states, and that mental states are pretty much what philosophers like Descartes take them to be: The capacity of speech acts to represent objects
and states of aairs in the world is an extension of the more biologically fundamental
capacities of the mind (or brain) to relate the organism to the world by way of such
mental states as belief and desire, and especially thought and action (1983: vii). And
scholars like Haugeland (1997) and Brandom (1994) will argue that the intentionality
of mental states is derivative of the intentionality of speech acts and, more generally,
that mental states and even speech acts are not what they are traditionally understood to be.
4. Speech acts were treated in The semiotic stance. Here the emphasis is on declarative
utterances.
5. Such an approach radically alters attempts to come to grips with intentionality in terms
of binary distinctions like originary versus derived, or speech act versus mental state,
and such an approach necessarily takes into account empirical research in primatology,
developmental psychology, anthropology, and linguistics.
6. Unlike Residence in the world, in which the organizing principle was dierent things
(the constituents of the residential whole) mediated in similar ways, the organizing
principle here is similar things (the constituents of the representational whole) mediated
in dierent ways.
7. For example, I know you are a husband insofar as 1) I saw the sign-event in which you
were married (e.g., a wedding), or 2) I see the patterns of interaction you have with
your spouse (e.g., exclusive lovemaking, shared credit cards, public intimacy, wedding
rings, and so forth).
8. Indeed, it was noted that, as a function of semiotic framing, embodied signs could be
understood in several ways: first, as an (ultimate) interpretant of another sign; second,
as a (dynamic) object that gives rise to a sign; third, as an (embodied) sign that gives
rise to an interpretant.
9. That is, the various antecedent sign events need not have any sensible properties in
common; and the various consequent sign events need not have any sensible properties
in common.
10. Though one suspects that the longer the latency, the more dicult the inference.
11. This is even true of purchases and functions. For example, from one particular interpretation of an artificed object (someone wields it a particular way), one can infer its
function, and then oer any number of other appropriate interpretants of it.
12. It should be noted that many scholars think our ability to understand psychological
forces comes before our ability to understand physical forces. For example, Collingwood argues that, Causal propositions . . . are descriptions of relations between natural
events in anthropomorphic terms (1972: 322, quoted in Tomasello and Call 1997: 388).
13. Joint attention (in the wide sense of this term) turning to observe what another is
observing or ostending, or observing or ostending so another turns to observe is
the exemplar of this two-part process: interpreting others signs of embodied signs;

Representations of the world

14.

15.
16.

17.
18.
19.

119

signifying embodied signs for others interpretants. In joint attention, the intentional
role is either an observing position or an ostensive action; the intentional status is the
object of observation or ostension; and the interpretant is a change in observing position. Thus, in joint attention, the intentional status need not have propositional content
(though it can have propositional content for parents and older children); the ground,
or relation between the intentional role and intentional status, is maximally iconicindexical (rather than indexical-symbolic); and the intentional attitude is an energetic
interpretant (though it can be an ultimate (representational) interpretant for parents
and older children). As noted, joint attention is the developmental milestone: nonhuman primates never acquire the ability; and human primates only acquire it at nine
to twelve months of age. For later observational and ostensive behavior, the intentional
attitude can be an ultimate representational interpretant, the ground can be symbolic
(look! or get a load of him), and the intentional status can have propositional content.
Purposes were prior to purchases, functions, statuses, and values; epistemic commitments are prior to empirical and practical commitments; and, somewhat paradoxically,
purposes are prior to epistemic commitments and hence purposeful actions are prior
to assertions (which are, of course, purposeful speech actions!), but assertions are prior
to intentional actions and observations. The following hierarchy emerges. Assertions
are prior to observations and (intentional) actions; (purposeful) actions are prior to affordances, instruments, roles, and identities; and (purposeful) actions are prior to assertions. Regarding consciousness, we might distinguish between merely phanerons (sensations due to incoming sound waves and light waves), experience (sensations paired
with non-propositional semiotic objects), and perception (sensations paired with semiotic objects. And ditto, regarding behavior, we might distinguish between mere movement, purposeful action, and intentional action.
This typology only focuses on intentional roles (as signs) and intentional statuses (as
objects), ignoring intentional attitudes (as interpretants).
And finally, in the realm of relatively non-emblematic roles, it should be stressed that
any mode of comportment (i.e., heeding an aordance, wielding an instrument, undertaking an action, performing a role, or filling an identity) may follow from an intentional status as a consequent sign event (or lead to an intentional status as an antecedent sign event), and hence constitute an intentional role. For example, the action of
walking across a rotting bridge may index a belief that it will hold ones weight, whereas tiptoeing across it may index a fear that it wont. Roles and identities index beliefs,
desires, structures of feeling, and so forth. Roles index knowledge: being a lawyer, a
doctor, a dressmaker, a wine taster, and so forth. Many emotions are inferred by anothers heeding of aordances in non-canonical ways (interpreting alleys as hiding
places), or wielding instruments in non-canonical ways (holding a knife upside down
as if to stab rather than slice). These signs are maximally metonymic: they primarily
index other objects (instruments index functions, actions index purposes, roles index
statuses, etc.), but they come to index intentional statuses by being in frequent contiguity with them precisely by being one of the fruits of that intentional status as an embodied sign. The key point, then, is that the fruits of embodied signs the consequent
sign events are often the best signs, and hence intentional roles, of that embodied
sign, or intentional status.
Here the proposition is given a mental interpretation (as maximally subjective); and the
state of aairs is given a worldly interpretation (maximally objective).
Brandom calls these doxastic commitments.
These need not be lexicalized in any language because they already have the most ubiquitous signs: declarative sentences, directions of attention, and controlled behaviors.

120

P. Kockelman

20. There is of course non-licensed attribution, and non-attributed undertakings: we attribute intentional statuses to people who did not do something that would license it; and
we undertake intentional statuses that people fail to attribute. All this depends on local
understandings of what counts as good evidence for an intentional status. The DSM
IV, for example is one expert account of what licenses a psychiatrist to attribute an intentional status depression, anxiety, and so forth. And, of course, being attributed
such an intentional status, especially if the attributor is an expert, often leads to acknowledging it.
21. As intimated, if there are intentional statuses and intentional roles, then there must be
intentional attitudes or others interpretants of ones intentional status which arise
because of ones intentional role. In particular, most intentional attitudes are just ultimate (representational) interpretants of others or ones own intentional status: treating
others or oneself as if epistemically, empirically, or practically committed to some
proposition, and sanctioning behavior accordingly. That is, intentional statuses, like
objects more generally, are primarily instituted by others attitudes towards them, as
evinced by the sanctioning practices of a community, and as embodied in the dispositions of its members.
22. Epistemic, empirical, and practical commitments are inferentially articulated: if one is
epistemically, empirically, or practically committed to a proposition, one is epistemically, empirically, or practically committed to any other propositions which may be
logically derived from it. Brandom (1994: 168170) characterizes inferential articulation in terms of inheritance of commitments and entitlements. Committive inference is
being committed to one proposition as a consequence of being committed to another
proposition. Permissive inference is being entitled to one proposition as a consequence
of being entitled to another proposition. And incompatible inference is having ones entitlement to a proposition be precluded as a consequence of being committed to another proposition.
23. Sellars (1963) referred to actions as language-exit moves (where position within
language game responded to via nonlinguistic situation), and to observations as
language-entry moves (where nonlinguistic situation responded to be adoption of position within language game), and to assertions or claims as intra-language moves
(where position in language game responded to be adoption of another position in language game).
24. This is Sellars interpretation of Kants maxim that percepts without concepts are
blind.
25. Loosely speaking, having an epistemic commitment caused by a state of aairs is the
rationalist version of Piageian accommodation (or Austins appropriateness); and having a practical commitment cause a state of aairs is the rationalist version of Piagian
assimilation (or Austins eectiveness). That is, perceptions are causally appropriate
and logically eective; intentions are logically appropriate and causally eective.
26. In Brandoms framework (1994: 261), intention is sometimes understood as an intentional status (practical commitment) and sometimes understood as an intentional attitude (acknowledgment of practical commitment). When they are understood as attitudes, intentions are causes: for in a properly trained agent acknowledgment of
practical commitment reliable causally elicits performances. This then is inferentialisms answer to the question of how mind aects body.
27. In the tradition of Anscombe (1957), Davidson (1980) argues that a performance (or
controlled behavior) is an action (under any description) if it is intentional under some
description. He argues that it is intentional under some description if that description
figures as the conclusion of a piece of practical reasoning that exhibits the agents

Representations of the world

28.

29.

30.

31.

32.
33.

34.
35.

36.

121

reasons for doing it. And he oers an account of primary reasons (e.g., a belief plus a
pro-attitude) to explain what it is for a reason to rationalize a controlled behavior according to a practical inference. Phrasing this in the idiom introduced here, a controlled behavior is intentional (under some description) and an action (under any description) insofar as it is the acknowledgment or self-attribution of a practical
commitment. And an intentional action is rational insofar as it stands as the conclusion
of an inference turning on an epistemic commitment and a pro-attitude (Brandom
1994: 233, 255).
Even Freud was an attempt to use desire to rationalize (unintentional) action. That is,
just as (saited) desire may be used to rationalize intentional actions, (frustrated) desires
may be used to rationalize unintentional actions.
To make this point explicit, note the following example: to attribute to an individual a
preference for staying dry is just to take inferences of this form (only remaining in the
car will keep me dry, so I will remain in the car), as entitlement preserving; and is to
license pattern of inferences. And to endorse such a pattern is to implicitly attribute
preference to individual that could be explicitly attributed by undertaking commitment
to ascriptional claim: A wants to stay dry (Brandom 1994: 248).
Crucially, Davidson (1980) thinks reasons are causes: primary reasons rationalize actions, first by providing reasons for them and second by serving to bring them about.
In contrast, Brandom thinks Davidson conflated commitment and entitlement to that
commitment: one can act intentionally without having reasons to do so; and one may
have a reason to act, and have an intention to act, but not act. For Brandom, the
causal nature of statuses happens because of attitudes towards statues insofar as statuses determine what roles are proper. In this way, the only access statuses have to the
causal order is through the attitudes of signers (1994: 260). What observable states of
aairs causally elicits in perception, is attitudes acknowledgments of empirical commitments; and what attitudes (acknowledgment of practical commitments) causally elicits in action is the production of performance. This is what it means to say that a performance must not only be caused by an intention, but be caused by it in the right way.
In sum, exhibiting a bit of practical reasoning rationalizes practical commitment; accepting a practical inference as entitling someone to practical commitment requires
endorsing inference as permissively good (but can be defeased by incompatible commitments), but doesnt require that scorekeeper endorse premise (which would pick
out objective entitlement or unconditional ought) (Brandom 1994: 253).
In general, there can be failures of presupposition (regarding the existence of a referent)
and failures of foci (regarding the applicability of predicate to a referent).
As Brandom notes, Thought of in this way, the distinction between de dicto and de re
should not be understood to distinguish two kinds of belief or even belief-contents, but
two kinds of ascription in particular two dierent styles in which the content of the
commitment ascribed can be specified (1994: 503).
In certain cases say, via reported speech this can happen: she said ouch!
As Brandom puts it, The introduction of a sentential operator that functions as S believes that . . . or S is committed to the claim that . . . does in English make it possible not merely implicitly or in practice to take someone to be committed to a claim, but
explicitly to say that someone is committed to a claim, and to which claim. The explicit
is the claimable, what can be given as a reason and have reasons demanded for it; ascriptional locutions make implicit attributions explicit as the contents of claims (1994:
498).
Finally, we might ask what are some of the key topics, questions, stakes, theories,
methodologies, methods, epistemic practices, and ethical practices of an episteme, and

122

37.

38.

39.

40.

41.
42.

P. Kockelman
what have they been in its past? These questions can, of course, be asked of any episteme; and part of knowing the history and culture of an episteme is knowing how the
answers to these questions, normatively speaking, change or remain the same over its
lifetime. Indeed, just as a large part of doing convincing research is knowing which of
these norms to adhere to, a large part of doing creative research is knowing which of
these norms to break.
Though, it may also be related to Wittgensteins form of life, Kuhns paradigms, Canguilhems history of epistemology, and latter-day science studies scholars who are immersed in and reacting to these ideas: Latour, Simon, Shan, Woolgar, and so forth.
This would involve related ideas such as the historical ontology of Hacking (2002);
though Hacking tends to focus on theory and intervention (see, for example, his
1983 monograph entitled Representing and Intervening).
Experiments with American and European middle- and upper-class children has offered a fairly consistent model of Western folk psychology. For example, there are
systematic taxonomic and partonomic interrelations among various mental states
(DAndrade 1995; Rips and Conrad 1989; and Wellman 1990). There is a notion of
the mind as distinct from the body, yet held in the brain and equivalent to the self
(Johnson 1987). There is a notion of the privateness of mental states, and their representational capacity (DAndrade 1995; Wellman 1990). There is a notion of real entities able to be distinguished from mental entities on the basis of sensory evidence, public existence, and temporal consistency (Wellman and Estes 1986). Some studies show
that there is a tendency to personify the mind, such that children move towards a conception of the mind as an independent entity (Wellman and Hickling 1994). And studies
show that subjects think that people can and should know the mental states of others.
These studies accord with the number of mental state terms in English there are over
200 word devoted to the emotions alone (Wallace and Carson 1973) and with the
propensity to use such terms in describing the behavior of others (Friestad and Wright
1995).
It should be stressed that one cannot account for emotions without simultaneously offering an account of selfhood and agency the former, as what is at stake in an interpretation; and the second as determining emotions as those mental entities over which
one has relatively little agency, as the western ethnotheory often has it.)
Indeed, most sophisticated accounts of emotions do not see them as internal states,
but as the relatively systematic bundling of some combination of the following components: eliciting situation; physiological change; reflexive signal; relatively controlled response; subjective feeling; and second-order interpretations of this ensemble of components as relatively uncontrollable, subjective, and natural. As may be seen with this
example, The semiotic stance naturally incorporates these components, and goes far
beyond them. Notice how the attempt to include appraisals in understandings of emotion is usually theorized as a way of bring concepts, or cognition, back into our understanding of aective phenomena. Notice that only representational interpretants require concepts per se; yet all interpretants require meaningfulness.
Natural kinds are categories that supposedly correspond to some real distinctions in
nature and around which theories are constructed (Griths 1997: 171).
Note that while aect programs may be cross-culturally shared, in dierent cultures
they may nonetheless have distinct behavioral entailments, co-occurring signs, modes
of interpretation, and classes of eliciting objects. In sum, while aect programs themselves may very well be natural kinds, they at best serve as a cross-culturally distributed
set of stereotypic physiological responses that individual cultures may experience, elaborate, and interpret in their own locally specific ways. To paraphrase what Sahlins

Representations of the world

123

(1977) said regarding Berlin and Kays (1969) seminal work on basic color terms, such
programs are not the imperatives of culture, but its implements.
43. Though perhaps for the purposes of local reasoning, a point not considered by such
authors.

References
Anscombe, G. E. M. (1957). Intention. Oxford: Blackwell.
Austin, J. L. (2003 [1955]). How to Do Things with Words, 2nd ed. Cambridge, MA: Harvard University Press.
Averill, James R. (1980). A constructivist view of emotion. In Emotion: Theory, Research,
and Experience, R. Plutchik and H. Kellerman (eds.), 305340. New York: Academic
Press.
Berlin, Brent and Kay, Paul (1969). Basic Color Terms: Their Universality and Evolution.
Berkeley/Los Angeles: University of California Press.
Brandom, Robert (1994). Making it Explicit: Reasoning, Representing, and Discursive Commitment. Cambridge, MA: Harvard University Press.
Brentano, Franz (1995 [1874]). Psychology from an Empirical Standpoint. New York: Routledge.
Colapietro, Vincent M. (1989). Peirces Approach to the Self: A Semiotic Perspective on
Human Subjectivity. Albany: Albany State University of New York Press.
Collingwood, R. J. (1972). Essay on Metaphysics. Chicago: Gateway.
DAndrade, Roy (1995). The Development of Cognitive Anthropology. Cambridge: Cambridge University Press.
Darwin, Charles (1965 [1872]). The Expression of the Emotions in Man and Animals. Chicago: University of Chicago Press.
Davidson, Donald (1980). Actions, reasons, and causes. In Essays on Actions and Events, 3
20. Oxford: Oxford University Press.
Dennett, Daniel C. (1978). Intentional systems. In Brainstorms: Philosophical Essays on
Mind and Psychology, 322. Montgomery/Vermont: Bradford Books.
Dummett, Michael (1994). The legacy of Brentano. In Origins of Analytic Philosophy, 28
42. Cambridge: Harvard University Press.
Ekman, Paul (1982). Emotions in the Human Face. Cambridge: Cambridge University Press.
Frank, R. H. (1988). Passions within Reason: The Strategic Role of the Emotions. New York:
Norton.
Friestad, M. and Wright, P. (1995). Persuasion knowledge: Lay peoples and researchers beliefs about the psychology of advertising. Journal of Consumer Research 22, 6274.
Givon, Talmy (1980). The binding hierarchy and the typology of complements. Studies in
Language 4 (3), 333377.
Goman, Erving (1981 [1979]). Footing. In Forms of Talk, 124159. Philadelphia: University of Pennsylvania Press.
Grice, Paul (1989). The causal theory of perception. In Studies in the Ways of Words, 224
247. Cambridge: Harvard University Press.
Griths, Paul E. (1997). What Emotions Really Are. Chicago: University of Chicago Press.
Hacking, Ian (1983). Representing and Intervening: Introductory Topics in the Philosophy of
Natural Science. Cambridge: Cambridge University Press.
(2002). Making up people. In Historical Ontology, 99114. Cambridge: Harvard University Press.

124

P. Kockelman

Haugeland, John (1998). The intentionality all-stars. In Having Thought: Essays in the Metaphysics of Mind, 127170. Cambridge: Harvard University Press.
Jakobson, Roman (1990). Shifters and verbal categories. In On Language, L. R. Waugh and
M. Monville-Burston (eds.), 386392. Cambridge: Harvard University Press.
James, William (1950 [1890]). The scope of psychology. In The Principles of Psychology, 1
11. New York: Dover.
Johnson, Mark (1987). The Body in the Mind. Chicago: University of Chicago Press.
Kockelman, Paul (2002). Stance and subjectivity among the Qeqchi-Maya: Minding language and measuring labor under neoliberal globalization. Unpublished PhD dissertation,
University of Chicago.
(2003a). The meaning of interjections in Qeqchi-Maya: From emotive reaction to social
and discursive action. Current Anthropology 44 (4), 467490.
(2003b). The interclausal relations hierarchy in Qeqchi-Maya. International Journal of
American Linguistics 69 (1), 2548.
(2005a). The semiotic stance. Semiotica 157 (1/4), 233304.
(2005b). Stance and subjectivity. Journal of Linguistic Anthropology 14 (2), 127150.
Lillard, A. (1998). Ethnopsychologies: Cultural variations in theories of mind. Psychological
Bulletin 123 (1), 332.
Premack, D. and Woodru, G. (1978). Does the chimpanzee have a theory of mind? Behavioral and Brain Sciences 4, 515526.
Rips, L. J. and Conrad, F. G. (1989). Folk psychology of mental activities. Psychological
Review 96 (2), 187207.
Ryle, Gilbert (1984 [1949]). The Concept of Mind. Chicago: University of Chicago Press.
Sahlins, Marshall (1977). Colors and cultures. In Symbolic Anthropology, J. L. Dolgin (ed.),
165180. New York: Columbia University Press.
Searle, John R. (1983). Intentionality: An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press.
Sellars, Wilfrid (1963). Science, Perception, and Reality. London: Routledge and Kegan
Paul.
(1997 [1956]). Empiricism and the Philosophy of Mind. Cambridge: Harvard University
Press.
Silverstein, Michael (1981). The limits of awareness ( Sociolinguistic Working Paper 84).
Austin: Southwest Educational Development Laboratory.
(1993). Of nominatives and datives: Universal grammar from the bottom up. In Advances
in Role and Reference Grammar, Robert D. Van Valin (ed.), 465498. Amsterdam/Philadelphia: John Benjamins.
Tomasello, Michael (1999). The Cultural Origins of Human Cognition. Cambridge, MA:
Harvard University Press.
Tomasello, Michael and Call, Josep (1997). Primate Cognition. New York: Oxford University Press.
Tomasello, Michael, Call, Josep, and Hare, Brian (2003). Chimpanzees understand psychological states the question is which ones and to what extent. Trends in Cognitive Science
7 (4), 153156.
Van Valin, Robert D. Jr. and LaPolla, Randy J. (1997). Syntax: Structure, Meaning, and
Function. Cambridge: Cambridge University Press.
Wallace, A. F. and Carson, M. T. (1973). Sharing and diversity in emotion terminology.
Ethos 1, 129.
Wellman, H. M. (1990). The Childs Theory of Mind. Cambridge, MA: Bradford Books.
Wellman, H. M. and Estes, D. (1986). Early understanding of mental entities: A reexamination of childhood realism. Child Development 57, 910923.

Representations of the world

125

Wellman, H. M. and Hickling, A. K. (1994). The minds I: Childrens conceptualization of

the mind as an active agent. Child Development 65, 15641580.
Whiten, A. (1993). Evolving a theory of mind: The nature of non-verbal mentalism in other
primates. In Understanding Other Minds, S. Baron-Cohen, H. Tager-Flusberg, and D. J.
Cohen (eds.), 367396. New York: Oxford University Press.
Whorf, Benjamin (1956 [1939]). The relation of habitual thought and behavior to language.
In Language, Thought and Reality, John B. Carroll (ed.), 134159. Cambridge, MA: MIT
Press.
Paul Kockelman (b. 1974) in an Assistant Professor in the Department of Anthropology,
Barnard College, Columbia University 3pk2113@columbia.edu4. His research interests include linguistic anthropology, social and semiotic mediation of subjectivity, and economy
and society. His major publications include The interclausal relations hierarchy in QeqchiMaya (2003); The meaning of interjections in Qeqchi-Maya: From emotive reaction to social and discursive action (2003); Stance and subjectivity (2004); and Psychological depth
is the internalization of dialogical breadth: Modal clitics and mental states in QeqchiMaya (2005).