Journal of Adult Development, Vol. 6, No.

1, 1999

Generation Identity and the Reminiscence Bump:

Memory for Public and Private Events

Alison Holmes,1 and Martin A. Conway1,2

Participants ranging in age from 30 to 70 years free-recalled when they had learned public
and private items of news. When responses were plotted in terms of age of participant at
time of encoding, it was found that peak recall for public items of news was in the period
when participants were aged 10 to 19 years whereas peak recall of private items of news
occurred in the period when participants were aged 20 to 29 years. A second study confirmed
this pattern of an early reminiscence bump for public news. It is proposed that these two
components of the reminiscence bump reflect, respectively, a period of formation of gen-
eration identity in the second decade of life and a period of formation of intimate relation-
ships in the third decade.
KEY WORDS: Memory for public events; autobiographical memory; generation identity; development
of the self; psychosocial stages.

INTRODUCTION as "among the most regular" of all memory phenom-

ena. In the same paper, however, Rubin further com-
In the lifespan retrieval curves of individuals over mented that the reminiscence bump should command
the age of about 35 years there is a marked increase our attention not simply because of its regularity but
in memories of events dating to when rememberers also because it quite clearly relates to a cluster of theo-
were 10 to 30 years of age and this is known as the retical issues of wide ranging significance (see Rubin
reminiscence bump (Rubin, Wetzler, & Nebes, 1986; et al., in press, for a recent review and evaluation of
see too Fitzgerald & Lawrence, 1984, and Franklin & these). Most importantly, it seems that the bump must
Holding, 1977, for earlier work on this phenomenon.). relate to the maturation of the self (Fitzgerald, 1988)
The reminiscence bump has been observed in a large which takes place during this period, 10 to 30 years
number of studies using free and cued recall with a of age, and which itself culminates in the emergence
large variety of populations (see Conway, 1996; Con- of a stable identity (Erikson, 1950). But what exactly
way & Rubin, 1993; Rubin, Rahhal, & Poon, in press; is the relation between the reminiscence bump and the
Rubin & Schulkind, 1997, for reviews). More recently developing self?
the reminiscence bump has also been observed for fac- Surprisingly, there are few well formed theoreti-
tual knowledge (Rubin et al., in press; Schuman, Belli, cal accounts of the development of self and even less
& Bischoping, 1997; Schuman & Scott, 1989). In fact
empirical research into what is surely one of the most
the reminiscence bump is so robust and ubiquitous
important aspects of human cognition. The work of
that Rubin (in Conway & Rubin, 1993) referred to it
Erikson (1950, 1985) remains the main theoretical ac-
count of the development of the self across the the
lifespan and so it is to this theory we turn. Erikson
1
University of Bristol, Bristol BS8 1TN, England. (1985) described eight psychosocial stages of devel-
2
Correspondence should be directed to Martin A. Conway, Centre
for Learning and Memory, Department of Experimental Psychol-
opment that extend sequentially over the entire
ogy, University of Bristol, 8 Woodland Road, Bristol BS8 1TN, lifespan. These stages are not mutually exclusive and
England; e-mail: M.AConway@bristol.ac.uk. Erikson's proposal was only that at any given point

21
1068-0667/0100-0021$16.00/0 C 1999 Plenum Publishing Corporation
22 Holmes and Conway

in the lifespan an individual will be primarily engaged groups and society as a whole) then it is privileged
in the resolution of one of these stages. Issues of self retention of knowledge of public rather than private
development relating to past stages, and possibly fu- experiences that may dominate during this stage.
ture stages, may also be present but not dominant The second and later psychosocial stage covered
during the resolution of a specific stage. The stages by the reminiscence bump is that of intimacy versus
which concern us here are those roughly correspond- isolation. The existential task of this stage is the for-
ing to the period of the reminiscence bump, adoles- mation of intimate personal relations and especially
cence and young adulthood, which according to a sexual relationship with a loved other with whom
Erikson's clinical and cross-cultural observations focus life tasks are to be shared in the context of a lasting
on the problems of identity formation versus identity and intimate friendship. Because the focus of this pe-
confusion, and intimacy versus isolation, respectively. riod is on adult relations with other (young) adults,
The period of identity formation is especially signifi- it should be the case that those experiences which
cant, for this is the time when the individual estab- relate most strongly to the goals of attaining intimacy
lishes personal goals that, in one form or another, are encoded in a way that renders them highly ac-
may endure for many years, even across the whole cessible. Here too these early and significant experi-
lifespan, as an integral self is consolidated and this is ences of intimacy might form the basis in long-term
a self that is more or less integrated with its imme- memory of organizational schemes that will later pro-
diate social groups and with society generally. Linking vide the indices for the encoding of future experi-
this stage to Mannheim's (1952) important concept ences of personal relationships. Given this line of
of generation identity (Conway, 1997; Schuman et al., argument, we would then expect this phase to be
1997), we might say that one of the important tasks marked by the retention of autobiographical memo-
of the Eriksonian psychosocial identity stage is the ries of interactions with significant others rather than
formation by the individual of a generation identity. the retention of experiences relating to groups and
Generation identity occurs when the individual society. In other words, this later period of the remi-
recognizes that he or she is part of a particular social niscence bump might show preferential retention of
subgroup with whom he/she shares common goals, private as opposed to public knowledge.
existential problems, knowledge, and, eventually, The view we have developed above proposes that
memories of experiences of a similar type (cf. Con- the reminiscence bump is the result of preferential or
way, 1997). For example, an individual in this stage privileged encoding of experiences relating to two
might identify with a particular social movement, po- crucial phases of development of the self in adoles-
litical group, religious group, or other subgroup of cence and early adulthood. Our reasoning leads us to
the time. It seems to us that there are several out- conjecture that retention of public knowledge might
comes of this for memory, one of which relates to be most marked in the early period of the reminis-
cence bump (10 to 20 years), whereas the retention
the amount of cognitive effort expended by the in-
of autobiographical memories of private experiences
dividual as the person oscillates between identify for-
might be more marked in the later part of the bump
mation and identity confusion with regressions back
(20 to 30 years). Relatedly, we also propose that the
to earlier modes of cognizing and progressions to
content of what is retained of public knowledge will
stages yet to be attained. Because of this investment
differ for different generations. We now describe two
of cognitive effort, knowledge is likely to be encoded experiments that investigated these hypotheses.
into memory in a privileged way compared to knowl-
edge encoded at later points in the life cycle. More-
over, knowledge retained during the period of
EXPERIMENT 1: RECALL OF PUBLIC AND
generation identity formation may form the basis of
PRIVATE EVENTS
autobiographical memory knowledge structures that
persist and which will be added to over the entire
lifespan. As the progenitors of these knowledge Method
structures, memories acquired during this stage may
remain highly accessible. Finally, we note that as Participants
identity formation, and generation identity formation
in particular, concern the internalization of selected One hundred adults aged 30 to 70 years, sepa-
features of the external world (immediate social rated into four groups of 25 by age, 30 to 39 year
 Generation Identity and Reminiscence 23

olds (mean 35 years); 40 to 49 years (mean 45 years); Results

50 to 59 (mean 55 years); and, 60 to 70 (mean 64
years), were recruited through advertisements in lo- In total, 2,514 dated events were collected
cal newspapers and billboards posted at Bristol Uni- (mean 25), with significantly more private than public
versity. Participants were paid £5.00. events listed, t = 2, p < .05. The percentage of
events, both public and private, falling in each of
seven decades was calculated and is plotted in Fig.
Design 1, shown by the unbroken line marked "All Events."
The lifespan retrieval curve for All Events in Fig. 1
shows a powerful reminiscence effect and 53.1% of
A within-subjects design was used in which par-
events dated to the second and third decades, when
ticipants freely listed both public and private events. participants were aged 10 to 29 years. The data were
Events was a repeated measure with two levels, pub- then divided into public and private events, and the
lic and private. All events were dated and the age of curves for these are also shown in Fig. 1. Contrasting
the participant at encoding was calculated. Age-at- the curve for Public Events with that for Private
encoding was the dependent measure. Responses Events, it can be seen that the reminiscence bump
were also categorised and category of response was peak for public events was in the 10- to 19-year dec-
a secondary dependent measure. ade whereas for private events it was in the 20- to
29-year decade. In order to make a formal contrast
here, the data were entered into an analysis of vari-
Procedure ance (ANOVA) with event type a within-subjects
variable with two levels (private and public) and
lifespan a within-subjects variable with four levels
Participants wrote their responses in experimen-
(first four decades, 0 to 40 years). Only the first four
tal booklets. They were instructed to list those public
decades were used, as only these decades featured
events occurring in their lifetime which they consid- events from all participants. The important finding
ered to be important. These could be local, national, from the ANOVA was a reliable Event Type x
or international events, the only constraint being that Lifespan interaction, F(9, 288) = 2.1, p < .04. Least-
they should list as many as possible in the 10-min squares means contrasts between decades for each
period available. They were asked not to edit events set of events separately found that for public events
but simply list them as they came to mind. For each reliably more events dated to the 10-19 decade
event a short title was required that would provide whereas for private events reliably more events dated
sufficient information for another person to be able to the 20-29 decade. This finding lends support to
to identify the event listed. Participants were further our earlier suggestion that the reminiscence bump
instructed to list private events from their own lives may be influenced by two different stages of self-de-
which they considered to be important. In this case velopment: an early stage in which the self integrates
they were to provide a title for each memory that with the external world and the focus is upon the
would indicate to a reader what the main content of retention of public events, and a later stage in which
the memory was. Presentation of tasks was counter- the self integrates with specific individuals and the
focus is upon private events.
balanced. Participants were tested in small groups
These findings are, however, complicated by a
and given written instructions followed by a verbal
further factor, namely the age of the participants.
summary of the tasks. After 10 min the experimenter
When the data were plotted for each age group sepa-
brought the session to a halt and participants then rately, somewhat different lifespan retrieval curves
undertook several unrelated tasks before, finally, were observed. Consider Fig. 2, which shows the
completing the remaining listing task. After this the curves for public events for each of the four age
experimental booklets were returned to the partici- groups. It can be seen from Fig. 2 that only the 40-
pants, who dated all responses. Public events were to 49- and 60- to 69-year-old groups showed the remi-
dated in terms of when the participants judged they niscence peak in the 10-19 decade. The lack of an
had first become aware of them and private events effect for the 30- to 39-year old group is, perhaps,
in terms of when they had occurred. somewhat misleading and may mask the expected ef-
24 Holmes and Conway

Fig. 1. Lifespan retrieval curves for memories of public and private events.

Fig. 2. Lifespan retrieval curves for memories of public events for different age groups.
 Generation Identity and Reminiscence 25

Fig. 3. Lifespan retrieval curves for memories of private events for different age groups.

fect with a recency component. The lack of effect for inequalities in classification, various categories were
the 50- to 59-year-old group is less understandable merged to produce five categories in each set of re-
but may have arisen because of low numbers of par- sponses, although even this led to imbalances. For
ticipants in these age groups (there were only 25 in public events, the categories were Family (Royal),
each group) and this may have given rise to some in- 9%; Political, 39%; War/Murder (assassination),
stability and changes due to inadvertent biases in sam- 38%; Sports/Ents. (Entertainments), 10%; and News
pling. The outcome is altogether more positive for Events, 16%. The Family (Royal) category comprised
Private Events x Age Group, and it can be seen from news of the marriages, divorces, births, deaths, anni-
Fig. 3 that all groups showed a marked reminiscence versaries, and scandals of prominent public figures
effect with peaks in the 20-29 decade (details of the and especially of the British Royal Family. The Po-
analyses of these data can be found in Holmes, 1999). litical category encompassed major political events
In order to further test our suggestion that the such as resignations, elections, scandals, etc. The
early phase of the reminiscence bump is influenced War/Murder category included acts of terrorism,
by the preferential encoding of public events whereas events from various wars, famous murders, and as-
the later phase is influenced by the preferential en- sassinations. Sports/Ents. was made up of mentions
coding of private experiences, we classified all the re- of famous sporting events and other public entertain-
sponses for public and private events separately. In ments, e.g., rock concerts, cup finals, etc. The News
both cases up to 10 categories were found to produce Events category included all other responses, and
the finest classification and account for over 95% of these mentioned disasters, natural and human made,
responses in both sets of events. However, there were as well as some other events not easily classified, e.g.,
striking imbalances between classes. For example, in companies going broke, weather phenomena, etc.
public events the category "Space" events (particu- For private events the five categories were Re-
larly the moon landing) accounted for only 3% of all lationship, 17%; Births/Deaths, 20%; Work/Educ.
responses whereas "political" events accounted for (Education), 29%; Home/Leisure, 24%; and Ill-
over 30%. In order to reduce at least some of these ness/Relig. (Religion), 11%. The Relationship cate-
26 Holmes and Conway

Fig. 4. Lifespan retrieval curves for memories of different classes of public events. Ents. = Entertainments.

gory contained mentions of marriages, divorces, and 10-19 and 20-29 decades, whereas other news events
other relationship events all closely linked to or a di- had their peak in the 30-39 decade.
rect part of the respondents own life. Births/Deaths Contrast this with Fig. 5 and the curves for the
again referred to these events in the respondent's five categories of private events. Here only the
own family and circle of friends. Work/Educ. com- Work/Educ. category has a peak in the 10-19 decade
prised mentions of events from the respondent's own and this is largely because of the recall of school
job, school days, university, education of own chil- memories from this period. The Relationship,
dren, etc. Home/Leisure referred to experiences in Births/Deaths, and Home/Leisure all peak in the 20-
the home, at dinner parties, and regarding home im- 29 decade, and especially striking here is the peak
for mentions of relationship experiences (see Fig. 5).
provements, moving home, etc. Finally, Illness/Relig.
The Illness/Relig. category peaks in the 30-39 and 40-
was dominated by mentions of illness and a few men-
49 decades, and this effect is largely for memories
tions of religious experiences as well as some unclass-
of periods of illness. Public events about well-known
ifiable responses. The percentages of responses public figures, news of events from wars, acts of ter-
across decades within categories for public and pri- rorism, murders, and assassinations were, then, found
vate events was then plotted and the resulting curves to date to the first decade of the bump. Retention
are shown in Figs. 4 and 5, respectively. of these news events may reflect individuals in the
Consider first Fig. 4, which shows the lifespan re- process of identifying with the zeigeist or spirit of
trieval curves for the different classes of public events. their times. In contrast, the later occurring retention
In can be seen in Fig. 4 that the peak point of recall of private memories of experiences from (adult) re-
for public events of royal family-associated experi- lationships was most clearly present in the second
ences, wars, murders, and acts of terrorism occurred decade of the bump, and this may reflect the indi-
in the 10-19 decade. Political events and sports/enter- vidual achieving intimacy at this stage of develop-
tainment events were equally frequent over both the ment. These findings lend further support to our
 Generation Identity and Reminiscence 27

Fig. 5. Lifespan retrieval curves for memories of different classes of private events. Educ. = Education; Relig. = Religion.

proposal that two different phases of development ences were also present in the 20-29 decade for public
of the self underlie the reminiscence bump. How- events. The 30- to 39-year-olds recalled most leisure-
ever, this was not the case for all categories of events, related events; the 40- to 49-year-olds recalled most
and, most noticeably, memories for public political marriage/divorce, home, and illness events; and the
events was equally high in both decades of the bump. 50- to 59-year-olds' and the 60- to 69-year-olds'
This latter finding may, perhaps, reflect a constant memories from this period were dominated by events
interest in political events from about the age of 10 that had featured births or deaths. A distinct impres-
years up to the age of 30 years. sion here is that the psychosocial stage of a remem-
Finally, and briefly, consider differences in fre- berer biases recall to events related to that stage the
quency of categories of events between age groups. rememberer is currently in. This is within a more gen-
For public events in the 10-19 decade 30- to 39-year- eral set of biases that originally influenced encoding
olds predominately mentioned political events and and, hence, the accessibility (organization) in memory
births/deaths events; 40- to 49-year-olds frequently of events originally closely related to the goals of the
mentioned political events but also murder/assassina- self in earlier psychosocial stages. We do not pursue
tions and various disasters; 50- to 59-year-olds men- these intriguing findings further here and a full analy-
tioned political events, marriages and divorces, and sis is available in Holmes (1999).
events related to public service; and 60- to 69-year-
olds most frequently mentioned war/terrorism events
and royal anniversaries. The overall differences of 60- EXPERIMENT 2: COMPLETING AMBIGUOUS
to 69-year-olds from the other groups was a World NAMES
War II effect, and their frequent mentions of royal
anniversaries may also have reflected a more powerful In Experiment 1 we found evidence indicating
identification with state and nation than that shown that the early part of the reminiscence bump might
in the memories of other groups. Age group differ- be predominately influenced by internalization of
28 Holmes and Conway

public occurrences while the latter part might be ambiguous single names. The ambiguous names were
dominated by internalization of private intimate ex- Charles, Stewart, James, Graham, Montgomery,
periences. In this second experiment, we focused on George, Ruth, Neville, Howard, Dean, Harrison,
the early part of the bump, the 10-19 decade, and Holly, Richard, Thomas, Joseph, Blake, Oswald, Dy-
examined differences between age groups in both lan, Scott, and Duncan. These were selected as they
type and content of knowledge recalled from this pe- were names which could have been either the first
riod. A further difference in this second experiment name or surname of many famous characters from
was that we wanted to avoid a direct instruction to the past 70 years. The unambiguous or less ambigu-
recall material. The reason for this was that we ous names were Botham, Blair, Trevor, Kinnock, Ma-
wanted to demonstrate that generation identity, as jor, Cantona, Tim, Redgrave, Anderson, Charlton,
reflected by privileged retention of public knowledge Linford, Connelly, Lenny, Wax, Linnekar, Neeson,
from the period when rememberers were aged 10 to Will, Nelson, Diamond, and Peter. The unambiguous
19 years, influences performance even in tasks that names were chosen because they were the names of
only indirectly require memory retrieval. In order to celebrities who have been featured widely in the me-
do this we developed an ambiguous names task in dia, both currently and in the past. These items were
which participants were presented with a name, e.g. included in order to disguise the purpose of the ex-
"John," and asked to supply either a forename or periment and also to ensure that names were not
surname that completed the name of a public figure.3 supplied only from very recent time periods.
Participants then described the first memory to come
to mind to this name or the memory that came to
mind when they generated it. Later the memories Procedure
were dated and plotted in lifespan retrieval curves.
Our predictions were that there would be a marked Instructions were provided on a typewritten
reminiscence peak in the 10-19 decade and that dif- cover sheet to the experimental booklet. Participants
ferent age groups would complete the ambiguous were instructed that the booklet contained names
names with names relevant to the public figures that which could either be first names or surnames, and
dominated their period of generation identity forma- that they should "read each one and complete it to
tion (cf. Sehulster, 1996). show the name of a famous person." They should
then write a short title describing the first memory
to come mind associated with the name. If a memory
Method had come to mind as they generated the name then
that was to be described as the associated memory.
Participants Participants were strictly instructed to respond with
the first name and first memory to come to mind.
One hundred adults aged 30 to 70 years (mean They were informed that the experimenter was only
age 50) participated and formed four groups of 25: interested in "snap" responses and that the purpose
30- to 39-year olds (mean age 36), 40- to 49-year- of the experiment was to examine the first ideas that
olds, (mean age 45), 50- to 59-year-olds (mean age came to mind during this task. Some examples of re-
54), and 60- to 70-year-olds, (mean age 64). Partici- sponses from an earlier pilot study were provided.
pants were paid a small honorarium. For instance, in response to the cue "Margaret" par-
ticipants had written: "Magaret Thatcher . . . former
Materials Prime Minister . . . saw her when she visited Mid-
dlesbrough"; "remember seeing her [Mrs. Thatcher]
Each participant was provided with an experi- on TV when she left 10 Downing Street; "Ann Mar-
mental booklet containing 20 ambiguous and 20 un- garet . . . remember her in the movie "Tommy" and
remember seeing Tommy at Odeon with Alice." Par-
3
If in the example you completed "John" with "John F. Kennedy," ticipants were instructed to work through the booklet
then according to our view you would have been about 15 years as quickly as they could trying not to miss any items.
old in 1963 when J.F.K. was assassinated, making you about 49 However, if nothing occurred within a few minutes,
(plus or minus 5) years old now. If, on the other hand, you com-
pleted it with, say, "John Travolta," a film star orginally famous they were to move on to the next item and return
in the mid-1970s, then you would now be approximately 37 years to the missed item when all the others had been com-
old, give or take 5 years. pleted. After completing the booklet, participants
 Generation Identity and Reminiscence 29

Fig. 6. Lifespan retrieval curves by age group for memories of famous names.

were then instructed to date each item by giving their

called in association with the generated famous
age in years and months at the time of the experience
names. Figure 6 shows the resulting lifespan curves.
that was recalled in regard to each name. The ex-
It can be seen from Fig. 6 that, first, there was no
periment lasted for approximately 1 hour.
recency effect and, second, that there was a large
reminiscence peak in the 10-19 decade and this was
Results the case for all groups. Overall, 34.3% of the asso-
ciated memories provided in response to self-gener-
The 20 ambiguous (target) cues generated 345 ated ambiguous names came from the 10- to 19-year
different names. In total 1,690 target items were in- decade. In order to analyze these data and the data
cluded in the data (mean response rate was 84.5%). from the responses to unambiguous names, repeated-
Unambiguous names generated 1,706 responses measures ANOVAs were conducted separately for
(mean response rate was 85.3%), but less than 30 each age group. In these ANOVAs, type of name
different names (confirming their lack of ambiguity). formed a variable with two Levels, ambiguous and
The distribution of response types to ambiguous unambiguous, and decade formed a second variable
name cues is shown in Table AI in the Appendix. It with Levels 4, 5, 6, and 7 for each age group, respec-
can be seen from Table AI that response frequencies tively. No significant effects of type of name were
ranged from a maximum of 28 for James, George, found in any age group, indicating that each group
and Richard at the upper end to three for Dylan responded in the same way to ambiguous and unam-
(mean = 17.25) at the lower end. Response rates biguous names. In each group there was, however, a
ranged from 92% for Charles to 73% for Blake. The significant Type of Name x Decade interaction: 30-
data for responses to ambiguous names only (unam- to 39-year-olds, F(3, 24) = 25.4, p < .01; 40- to 49-
biguous names had been included as a control and year-olds, F(4, 24) = 16.4, p < .01; 50- to 59-year-
showed a flat distribution over the lifespan) were olds, F(5, 24) = 9.5, p < .01, and 60- to 70-year-olds,
then tabulated and subsequently plotted in terms of F(6, 24) = 12.6, p < .01. These interactions occurred
age of the participant at the time of each event re- because the distribution for unambiguous names
30 Holmes and Conway

showed a slight recency effect, with a flat distribution tures of lifelong experiences of intimacy. The knowl-
across decades back to the 0-9 decade where a de- edge encoded in this latter part of the reminiscence
crease in responses was evident, and this contrasts period will, of course, be in part determined by what
markedly with the distribution of responses to am- has gone before. The unique feature at this point in
biguous names shown in Fig. 6. development is that intimate personal relationships
are formed outside the context of the family and so
are to at least some degree independent of direct in-
GENERAL DISCUSSION fluences of parental and other authority figures—the
formation and conduct of such relationships is the
responsibility of the newly independent self of the
We have proposed that the reminiscence bump
young adult, which also ensures good retention of
in the lifespan retrieval curve is a product of privi-
these formative experiences (see Conway & Pleydell-
leged encoding of experiences highly relevant to an
Pearce, 1998, and Robinson, 1992).
individual during a critical phase of development and
By our view, then, there are two major compo-
consolidation of the self. Our findings give good pre-
nents in the reminiscence bump. An early component
liminary support to this view and show that recall of
makes available memories relating to identification
knowledge of public events peaks during the period
with society and generational groups whereas a later
when rememberers were approximately 10 to 19
component makes available autobiographical knowl-
years of age whereas recall of knowledge of more
edge of formative adult personal relationships. Our
private or personal events tends to peak during the
data support this view but there are also other find-
period when rememberers were approximately 20 to
29 years of age. Why should this be? Following Erik- ings that add to the case. For example, an examina-
son (1985), we believe that the early period of the tion of the figures in Rubin et al. (in press) shows
reminiscence bump corresponds to a period when that the most important memories of private events
many individuals are undergoing a process of making have a peak when rememberers would have been
an external or generational identity with their society. about 23 to 25 years of age (see their Figs. 9 and
This entails identifying public events, beliefs systems, 10). In contrast, memories of public events of World
activities, etc., which are in some way appropriate to War II and the J.F.K. assassination tend to peak to-
the goals of the self during this period (cf. Conway, ward the end of the 10- to 20-year decade. Suggestive
1996; Conway & Plyedell-Pearce, 1998). These exter- though this may be, other distributions of memories
nal experiences receive privileged encoding by virtue for songs, entertainment, books, and films (Sehulster,
of being processed more deeply than other types of 1996) tend to have their peak reminiscence in the
experiences, both at the time of the experiences and 20- to 30-year decade. Thus, it cannot be the case
also, perhaps, in intense sessions of rehearsal when that all public events are retained because they were
the events were ruminated upon and discussed at related to identity formation and, in particular, for-
length. A further aspect of privilege encoding lies in mation of generation identity. Nevertheless, it may
the organization of knowledge in long-term memory be that newsworthy events and certain cultural events
(Conway, 1996), and public experiences occurring impinge more on the developing identities of 10- to
during this early part of the period of the reminis- 20-year-olds than upon members of other age groups
cence bump form the bases of long-term memory and hence upon their memorability reflected in the
knowledge structures which will later be used in the first half of the reminiscence bump.
comprehension and representation of subsequent as- These types of generational effects can also be
sociated public events (cf. Conway & Haque, 1998). detected in the content of memories from different
In a similar way the later period of the reminiscence age groups for the same period of the reminiscence
may correspond to a period when individuals are pri- bump. Thus, the event categories in Experiment 1
marily concerned with the development of intimacy hint at generational differences in memory content.
in personal relationships. As a consequence experi- For instance, political events dominated the public
ences relevant to intimacy and to the goals of the event type for all age groups except the 60- to 70-
self which detail the types of intimacy to be attained year-olds. This latter group showed a dominance of
receive a similar type of privileged encoding: The ex- memories of events associated with war and terror-
periences are processed to a deeper level, and form ism. Thirty- to thirty-nine-year-olds were the only
focal points in long-term memory knowledge struc- group to record Tianemen Square, the election of
 Generation Identity and Reminiscence 31

a female Prime Minister, decimalization, and the Striking generational differences were also ob-
Falklands and Gulf wars, although all older partici- served by Conway and Haque (1998), who found that
pants had lived through these events. This was also an older group of Bangladeshi people had a second
the only group not to mention the J.F.K. assassina- reminiscence bump in the period when they would
tion. In contrast, the most frequent response for 40- have been aged 35 to 55 years. The second bump con-
to 49-years-olds was the J.F.K. assassination, and tained many memories for events from the struggle for
this was the only group to mention the Aberfan dis- independence by Bangladesh from Pakistan which per-
aster and Korea. The 50- to 59-year-olds' events re- sisted over this lengthy period. It is inevitable that the
flected the beginning of a postwar era and the content of memories varies from individual to individ-
majority of this group mentioned events related ual and from generation to generation. However, what
either to the end of World War II and/or the coro- is important here is the type of memory that is recalled
nation of Queen Elizabeth II (in 1952). This group of which individual memories are tokens. Our sugges-
contained the only mention of the Festival of Brit- tion is that generational groups are bound together by
ain. First responses from the 60- to 69-year-olds sharing memories of a common type, i.e., memories
were dominated by World War II (52%); of these of a struggle for independence, war in Europe, war in
the majority mentioned the outbreak of World War Asia, relaxation of social prohibitions and changing so-
II, but only a single participant mentioned the end, cial values as in Western countries in the 1960s, and
and no other war was mentioned. The 30- to 39- so forth. Just as autobiographical memory appears to
year-olds to 40- to 49-year-olds were the only contain abstract or conceptual knowledge of lengthy
groups to mention the war in Vietnam in their first periods of time, so-called lifetime periods (Conway,
responses (see Holmes, 1999, for a more detailed 1996), e.g., "when I lived with X," "worked at company
account). Despite these differences, it is clear from Y," "lived in city Z," etc., it may also contain abstract
Table I that there were many public figures who or conceptual knowledge that contributes to identify-
spanned the generations (when these are crudely ing an individual's sociohistorical location. Sehulster
expressed as age groups as in Experiment 2). Inter- (1996) referred to this as "my ear," a period of time
estingly, although figures such as Prince Charles marked not only by its duration but also by it's char-
were mentioned equally frequently by participants acteristics and event content. The interface of self and
of different ages, it was often the case that different society in memory has yet to be investigated, and we
age groups recalled different events from the life offer our studies as one starting point. It seems to us
of the public figure mentioned. In the case of Char- that future studies focused in more detail on the na-
les, events relating to his schooling and childhood ture of identity formation during the 10- to 30-year-old
were recalled by the older groups, whereas for the period which collected fuller accounts of recalled pub-
younger group events relating to his marriage and lic events might be able to map interesting generation
subsequent separation were more frequently re- differences in some depth and, in so doing, develop a
called. Content difference in memories may then theoretical account of how memory helps determine
indicate subtle generation differences. social identity.

APPENDIX
Table AI. Distribution of Responses by Age Group to Ambiguous Names
Response 30- to 39-year-olds 40- to 49-year-olds 50- to 59-year-olds 60- to 70-year-olds
Cues types % (n = 25) (n = 25) (n = 25) (n = 25)
Charles 13 92 Prince Charles (68%) Prince Charles (56%) Prince Charles (56%) Prince Charles (64%)
Craig Charles (8%) Ray Charles (12%) Ray Charles (12%) Ray Charles (12%)
Others (16%) Charles de Gaulle (8%) Others (16%) Charles de Gaulle (8%)
Missing (8%) Others (16%) Missing (16%) Charles Boyer (8%)
Missing (8%) Others (8%)

Stewart 13 87 Rod Stewart (32%) Rod Stewart (36%) Rod Stewart (24%) James Stewart (44%)
Jackie Stewart (20%) Jackie Stewart (24%) Jackie Stewart (20%) Stewart Grainger (16%)
James Stewart (16%) James Stewart (16%) Stewart Grainger (20%) Others (28%)
Others (16%) Others (12%) Others (24%) Missing (12%)
Missing (16%) Missing (8%) Missing (12%)
32 Holmes and Conway

Table AI. Continued

Response 30- to 39-year-olds 40- to 49-year-olds 50- to 59-year-olds 60- to 70-year-olds
Cues types % (n = 25) (n = 25) (n = 25) (n = 25)
James 28 86 James Stewart (16%) James Stewart (20%) James Stewart (16%) Syd James (12%)
James Dean (12%) James Joyce (8%) James Mason (12%) James Hunt (12%)
Others (60%) Clive James (8%) Jesse James (8%) Henry James (12%)
Missing (12%) Others (48%) Henry James (8%) Others (56%)
Missing (16%) Others (36%) Missing (8%)
Missing (20%)

Graham 22 92 Graham Greene (24%) Billy Graham (28%) Graham Gooch (16%) Billy Graham (32%)
Graham Hill (12%) Graham Greene (16%) Graham Greene (12%) Others (56%)
Billy Graham (12%) Others (48%) Others (64%) Missing (12%)
Others (40%) Missing (8%) Missing (8%)
Missing (12%)

Mont- 7 82 Field Marshall Field Marshall Field Marshall Field Marshall

gomery Montgomery (28%) Montgomery (28%) Montgomery (48%) Montgomery (56%)
Montgomery Clift (24%) Montgomery Clift (24%) Montgomery Clift (36%) Montgomery Clift (24%)
Colin Montgomery (12%) Elizabeth Montgomery (8%) Missing (16%) Elizabeth Montgomery (8%)
Missing (36%) Others (8%) Others (12%)
Missing (32%)

George 28 89 George Best (12%) George Best (28%) Boy George (20%) Boy George (32%)
Boy George (12%) Boy George (20%) George Harrison (16%) George Harrison (12%)
George Michael (8%) George Michael (12%) Eddie George (12%) Others (32%)
Others (60%) Others (40%) George Best (12%) Missing (24%)
Missing (8%) Others (24%)
Missing (16%)

Ruth 13 79 Ruth Rendell (40%) Ruth Rendell (48%) Ruth Rendell (40%) Babe Ruth (36%)
Babe Ruth (16%) Babe Ruth (12%) Babe Ruth (16%) Ruth Rendell (20%)
Others (16%) Others (8%) Others (28%) —Old Testament (12%)
Missing (28%) Missing (32%) Missing (16%) Other (24%)
Missing (28%)

Neville 15 74 Neville Chamberlain (16%) Neville Chamberlain (52%) Neville Chamberlain (36%) Neville Chamberlain (60%)
Aaron Neville (12%) Neville Shute (8%) John Neville (16%) Neville Shute (16%)
Others (24%) Others (12%) Gary Neville (8%) John Neville (8%)
Missing (48%) Missing (28%) Others (16%) Others (16%)
Missing (24%)

Howard 16 87 Michael Howard (20%) Michael Howard (32%) Michael Howard (16%) Trevor Howard (28%)
Frankie Howard (12%) Frankie Howard (20%) Katherine Howard (16%) Howard Keel (20%)
Others (52%) Trevor Howard (16%) Howard Hughes (16%) Michael Howard (20%)
Missing (16%) Other (24%) Others (40%) Others (20%)
Missing (8%) Missing (12%) Missing (12%)

Dean 11 86 James Dean (56%) James Dean (36%) James Dean (68%) James Dean (44%)
Dean Stockwell (8%) Dean Martin (20%) Dean Martin (12%) Dean Martin (32%)
Dean Martin (8%) Others (24%) Other (4%) Dixie Dean (8%)
Others (12%) Missing (20%) Missing (16%) Others (12%)
Missing (16%) Missing (4%)

Harrison 7 94 Harrison Ford (76%) George Harrison (48%) Harrison Ford (44%) Harrison Ford (44%)
Others (12%) Harrison Ford (36%) George Harrison (36%) Rex Harrison (28%)
Missing (12%) Others (16%) Rex Harrison (12%) George Harrison (20%)
Missing (8%) Other (4%)
Missing (4%)
Generation Identity and Reminiscence 33

Table AI. Continued

Response 30- to 39-year-olds 40- to 49-year-olds 50- to 59-year-olds 60- to 70-year-olds
Cues types % (n = 25) (n = 25) (n = 25) (n = 25)

Holly 10 78 Buddy Holly (28%) Buddy Holly (52%) Buddy Holly (44%) Buddy Holly (64%)
Holly Hunter (24%) Holly Hunter (20%) Others (32%) Others (20%)
Others (16%) Others (12%) Missing (24%) Missing (16%)
Missing (32%) Missing (16%)
Richard 28 89 Richard Madeley (12%) Cliff Richard (24%) Cliff Richard (24%) Cliff Richard (24%)
Cliff Richard (12%) Richard III (20%) Richard III (16%) Richard III (16%)
Others (56%) Others (44%) Little Richard (12%) Little Richard (12%)
Missing (20%) Missing (12%) Others (44%) Others (44%)
Missing (4%) Missing (4%)
Thomas 22 89 —the Tank Engine (24%) —the Tank Engine (24%) Terry Thomas (28%) Thomas Hardy (20%)
Dylan Thomas (16%) Dylan Thomas (24%) Dylan Thomas (24%) Dylan Thomas (16%)
Others (40%) Terry Thomas (16%) —the Tank Engine (16%) Terry Thomas (12%)
Missing (20%) Others (28%) Others (28%) -the Tank Engine (12%)
Missing (8%) Missing (4%) Others (28%)
Missing (12%)
Joseph 19 81 Lesley Joseph (20%) Keith Joseph (28%) Keith Joseph (24%) Keith Joseph (24%)
-of Nazareth (12%) Joseph Cotton (12%) —Old Testament (12%) Joseph Cotton (16%)
Others (36%) Others (44%) -of Nazareth (12%) Joseph Conrad (12%)
Missing (32%) Missing (16%) Others (36%) Others (36%)
Missing (16%) Missing (12%)
Blake 14 73 Blake Edwards (24%) William Blake (28%) William Blake (28%) Sexton Blake (28%)
William Blake (16%) Blake 7(12%) George Blake (20%) William Blake (16%)
Others (16%) George Blake (12%) Sexton Blake (12%) Others (32%)
Missing (44%) Others (28%) Others (20%) Missing (24%)
Missing (20%) Missing (20%)

Oswald 4 84 Lee Harvey Oswald (48%) Lee Harvey Oswald (64%) Lee Harvey Oswald (40%) Oswald Mosely (76%)
Oswald Mosely (16%) Oswald Mosely (28%) Oswald Mosely (40%) Lee Harvey Oswald (16%)
Missing (36%) Missing (8%) Other (4%) Others (8%)
Missing (16%)

Dylan 3 90 Bob Dylan (60%) Bob Dylan (76%) Bob Dylan (64%) Dylan Thomas (68%)
Dylan Thomas (16%) Dylan Thomas (20%) Dylan Thomas (28%) Bob Dylan (24%)
Missing (24%) Missing (4%) Missing (8%) Other (4%)
Missing (4%)

Scott 19 82 —of the Antartic (28%) —of the Antartic (20%) Peter Scott (16%) —of the Antartic (32%)
Peter Scott (16%) Peter Scott (20%) —of the Antartic (12%) Peter Scott (32%)
Scott Fitzgerald (12%) Scott Walker (12%) Sir Walter Scott (12%) Others (28%)
Others (16%) Others (28%) Others (44%) Missing (8%)
Missing (28%) Missing (20%) Missing (16%)

Duncan 17 76 Duncan Goodhew (32%) Duncan Goodhew (32%) Duncan Goodhew (36%) Duncan Goodhew (28%)
Others (24%) Others (44%) Duncan Edwards (12%) King Duncan (12%)
Missing (44%) Missing (24%) Others (36%) Others (48%)
Missing (16%) Missing (12%)

ACKNOWLEDGMENTS sity of Bristol, and by a Economic and Social

Research Council award to Alison Holmes.
This research was supported by the De- The authors thank both institutions for their
partment of Experimental Psychology, Univer- support.
34 Holmes and Conway

REFERENCES Holmes, A. (1999). Patterns of reminiscence in autobiographi-

cal memory. Unpublished doctoral dissertation, Depart-
ment of Experimental Psychology, University of Bristol,
Conway, M. A. (1996). Autobiographical memories and autobio- Engalnd.
graphical knowledge. In D. C. Rubin (Ed.), Remembering our Mannheim, K. (1952). The problem of generations. In K.
past: Studies in autobiographical memory (pp. 67-93). Cam- Mannheim (Ed.), Essays on the sociology of knowledge (pp.
bridge, England: Cambridge University Press. 276-321). London, England: Routledge & Keegan Paul.
Conway, M. A. (1997). The Inventory of Experience: Memory and Robinson, J. A. (1992). First experience memories: Contexts and
identity. In D. Jodelet, J. Pennebaker, & D. Paez (Eds.), Po- function in personal histories. In M. A. Conway, D. C. Rubin,
litical events and collective memories (pp. 21-46). London, H. Spinnler, W. Wagenaar (Eds.), Theoretical perspectives on
England: Routledge. autobiographical memory (pp. 223-239). Dordrecht, The Neth-
Conway, M. A., & Haque, S. (1999). Overshadowing the reminis- erlands: Kluwer.
cence bump: Memories of a struggle for independence. Jour- Rubin, D. C, Wetzler, S. E., & Nebes, R. D. (1986). Autobio-
nal of Adult Development (this issue). graphical memory across the adult lifespan. In D. C. Rubin
Conway, M. A., & Pleydell-Pearce, C. W. (1998). On the construction (Ed.), Autobiographical Memory (pp. 202-221). Cambridge,
of autobiographical memories: The self-memory system and its England: Cambridge University Press.
neuroanatomical basis. Manuscript submitted for publication. Rubin, D. C., Rahhal, T. A., & Poon, L. W. (in press). Things
Conway, M. A., & Rubin, D. C. (1993). The structure of autobio- earned in early adulthood are remembered best. Memory &
graphical memory. In A. E. Collins, S. E. Gathercole, M. A. Cognition.
Conway, & P. E. M. Morris (Eds.), Theories of memory (pp. Rubin, D. C., & Schulkind, M. D. (1997). The distribution of auto-
103-137). Hove, Sussex, England: Lawrence. biographical memories across the lifespan. Memory & Cogni-
Erikson, E. H. (1950). Childhood and society. New York: W. W. tion, 25, 859-866.
Norton & Company. Schuman, H., Belli, R. F., & Bischoping, K. (1997). The genera-
Erikson, E. H. (1985). The life cycle completed. New York: Norton. tional basis of historical knowledge. In D. Jodelet, J. Penne-
Fitzgerald, J. M. (1988). Vivid memories and the reminiscence baker, & D. Paez (Eds.), Political events and collective
phenomenon: The role of a self narrative. Human Develop- memories (pp. 47-78). London, England: Routledge.
ment, 31, 261-273. Schuman, H., & Scott, J. (1989). Generations and collective
Fitzgerald, J. M., & Lawrence, R. (1984). Autobiographical mem- memories. American Sociological Review, 54, 359-381.
ory across the lifespan. Journal of Gerontology, 39, 692-698. Sehulster, J. R. (1996). In my era: Evidence for the perception of
Franklin, H. C, & Holding, D. H. (1977). Personal memories at a special period of the past. Memory, 4, 145-158.
different ages. Quarterly Journal of Experimental Psychology,
29, 527-532.