Journal of Vision (2016) 16(11):18, 1–4 1

Clarifying signal detection theoretic interpretations of the

Müller–Lyer and sound-induced flash illusions
University of California, Los Angeles,
J. D. Knotts Los Angeles, CA, USA $
University of California, Los Angeles,
Ladan Shams Los Angeles, CA, USA

In a recent article, Witt, Taylor, Sugovic, and Wixted

(2015) made several claims about the way the Müller– The Müller–Lyer illusion
Lyer and sound-induced flash illusions should influence
the signal detection theory criterion and sensitivity The centerpiece of Witt et al. (2015) is a computer
measures, c and d 0 . Here, we address some crucial simulation of a behavioral experiment using the
conceptual inconsistencies in their simulation of the Müller–Lyer illusion. In brief, the Müller–Lyer illusion
Müller–Lyer illusion and clarify a previous analysis of the occurs when adding inward-facing or outward-facing
sound-induced flash illusion from the literature that is tails to the ends of a horizontal line, which makes the
misinterpreted in their discussion. Alternative signal line appear shorter or longer respectively. On each trial,
detection theoretic interpretations of both illusions are a simulated observer is presented with either a short
offered. line (5 cm) or a long line (7 cm), with tails that are
oriented either inwardly or outwardly. The description
of the simulated effect of the illusion reads, ‘‘. . .on each
Introduction trial, the perceived line length was drawn from a
Gaussian distribution with the mean set to the sum of
the actual line length plus the theorized influence of the
A recent article by Witt, Taylor, Sugovic, and tails (þ13% for tails-out, 13% for tails-in)’’ (p. 4). The
Wixted (2015) reiterated and expanded on the impor- simulated observer then uses a fixed criterion at 6 cm to
tant observation that the signal detection theoretic perform the task of judging whether the line is long or
criterion measure, c, cannot distinguish between short. A hit occurs when the line is long (7 cm) and the
response biases and perceptual biases (Morgan, Hole, observer responds that the line is long (perceived line
& Glennerster, 1990; Raslear, 1985; Witt et al., 2015). length .6), and a false alarm occurs when the line is
They pointed out a pervasive misconception in the short (5 cm) and the observer responds that the line is
psychophysics literature that changes in the criterion long (perceived line length .6). The simulation is run
measure necessarily reflect changes in response bias, with 10 observers, each of whom performs 1,000 trials
and they used several well-known perceptual illusions (250 of each of four trial types: long tails-in, long tails-
to illustrate how such criterion shifts could instead be out, short tails-in, and short tails-out). Witt et al. (2015)
interpreted as representing perceptual bias. Their reported that when using the resulting simulated
manuscript thus serves as an important warning about behavioral data to compute the SDT measures d 0 and c,
the potential miscategorization of bias effects in signal they find a large change in c, but no change in d 0
detection theory (SDT) analyses. However, despite the between the tails-in and tails-out conditions. This, they
value in this overarching message, some of the argue, provides evidence for the fact that a perceptual
supporting arguments in Witt et al. (2015), particularly effect can be reflected in the criterion measure and not
those related to the Müller–Lyer and sound-induced in d 0 .
flash illusions, contain conceptual inconsistencies and The critical point to notice in the description of their
other points of confusion regarding the correspondence simulation that makes their reported result unexpected
between SDT measures and underlying perceptual is that the effect of the Müller–Lyer illusion is described
processes. We attempt to clarify these points below. as being modeled in accordance with the Weber–

Citation: Knotts, J. D., & Shams, L. (2016). Clarifying signal detection theoretic interpretations of the Müller–Lyer and sound-
induced flash illusions. Journal of Vision, 16(11):18, 1–4, doi:10.1167/16.11.18.
doi: 10 .116 7 /1 6. 11 .1 8 Received December 8, 2015; published September 30, 2016 ISSN 1534-7362

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.
Downloaded From: https://jov.arvojournals.org/ on 08/14/2018
 Journal of Vision (2016) 16(11):18, 1–4 Knotts & Shams 2

Figure 1. Distributions of perceived line length under different illusion conditions in the Müller–Lyer simulation described in Witt et al.
(2015). The distance between distribution means (vertical bars) changes with each condition. Tails-in and tails-out distributions are
shifted left and right, respectively, by 13% of the means of the corresponding no-tails distributions. All distributions have a standard
deviation of 1.2 cm. Note that the no-tails condition was not simulated in Witt et al. (2015), but is included here for illustration.

Fechner law, where the magnitude of the illusion is Based on the results above, we conclude that the
proportional to the size of the visual stimulus. This most likely explanation for the discrepancy between
relationship is intuitive and has been supported the reported simulation design and the results in Witt
empirically in humans (Tudusciac & Nieder, 2010). et al. (2015) is that the authors did indeed model the
However, if this is the case, then the distance between effect of the Müller–Lyer illusion by shifting the two
the means of the two perceived length distributions will evidence distributions by the same constant. It is
be different between the different illusory conditions unclear whether the mistake was in the reporting or
(Figure 1). If we then consider that all simulated
distributions have a constant standard deviation of 1.2
cm, we should expect the results of the simulation to
show a change in d 0 between the tails-in and tails-out
conditions. Indeed, when we ran this simulation, a two-
tailed, paired-samples t test showed a significant
difference in d 0 between these two conditions (d 0 tails-in
¼ 1.43 6 0.14, d 0 tails-out ¼ 2.02 6 0.18, t(9) ¼ 9.499, p
, 0.001; Figure 2A). This raises the question of where
the discrepancy lies between our simulation and that of
Witt et al. (2015). Given that they reported a constant
standard deviation for all simulated distributions, it
follows from SDT that the only way to maintain a
constant value of d 0 while changing the criterion is to
shift the two evidence distributions by the same
constant. We modified our simulation to do this by Figure 2. Results from the simulation of the Müller–Lyer illusion.
shifting the means of the tails-out and tails-in (A) Modeling the effect of the illusion by shifting the two
distributions relative to the distributions in the no-tails evidence distributions by the 13% of their respective means, as
condition by 613% of 6 respectively, and the resulting described in Witt et al. (2015). (B) Modeling the effect of the
SDT measures matched those reported in Witt et al. (d 0 illusion by shifting the two evidence distributions by a constant.
tails-in ¼ 1.81 6 0.13, d 0 tails-out ¼ 1.69 6 0.18, t(9) ¼ The results in (B) match those presented in Witt et al. (2015).
1.55, p ¼ 0.16; Figure 2B). *P , 0.001.

Downloaded From: https://jov.arvojournals.org/ on 08/14/2018

 Journal of Vision (2016) 16(11):18, 1–4 Knotts & Shams 3

the running of the simulation. In the former case, it is The real issue here is that we should expect different
left to be explained why the authors chose this less changes in SDT measures depending on which
intuitive model of the illusion, for which empirical conditions of the sound-induced flash illusion we
support in the literature is lacking. In the latter case, it choose to compare. This point is made clear by Figure
needs to be clarified that when the illusion is simulated 3, which provides an example of how d 0 and c change
as was reported, there in fact is a change in d 0 between between the three different auditory conditions (i.e.,
the tails-in and tails-out conditions. In either case, the zero beeps, one beep, and two beeps) in Rosenthal et
question is raised of whether or not the Müller–Lyer al. (2009). The fact that there is no significant change
illusion provides an appropriate example of an in d 0 between the two multisensory conditions does
experimental manipulation that can affect the criteri- not imply that the change in d 0 between unisensory
on without affecting d 0 . This is not to say that the and multisensory conditions is somehow not reflective
effect on the criterion should not be considered a of the sound-induced flash illusion. The two compar-
perceptual bias. We do not wish to undermine this isons are fundamentally different, and, in fact, it is
point, but rather to show that, in what may have been intuitive that d 0 should change between the latter and
haste to provide more convincing evidence for an not the former.
effect that supports the overarching message of their The purpose of comparing d 0 across the no-beep and
article, the authors have presented information that two-beep conditions in Rosenthal et al. (2009) was to
could potentially misguide future assumptions about track the magnitude of the perceptual component of
the way the perceptual processes underlying the the illusion that is due to multisensory integration. A
Müller–Lyer illusion are reflected in SDT measures. comparison between the one-beep and two-beeps
Given the pervasive use of this illusion in the study of conditions was not necessary for this purpose, and
visual perception, it seems crucial that this point be therefore it was left out of the analyses. This is not to
clarified. say that the change in c between the two multisensory
conditions cannot be reflective of perceptual process-
ing, but only that this particular measure was not of
primary interest to the goals of the study. We do agree,
The sound-induced flash illusion however, based on the central point of Witt et al.
(2015), that considering shifts in the criterion as
It is in the same vein that Witt et al. (2015) reflective of changes in perceptual processing could
oversimplified a signal detection theoretic explanation have potentially provided a more complete capture of
of the sound-induced flash illusion, and in doing so, the perceptual effect of the sound-induced flash illusion
misrepresented the SDT analysis provided in Rosen- between unisensory and multisensory conditions. Yet it
thal, Shimojo, and Shams (2009). Their discussion of stands to be corrected that, contrary to what was stated
the sound-induced flash illusion centers on a compar- in Witt et al. (2015), using d 0 to track perceptual effects
ison of the SDT measures computed for the one-beep of the sound-induced flash illusion is not, in and of
and two-beep conditions presented in Rosenthal et al. itself, an invalid method.
(2009). Comparing across these two conditions shows
exactly what the authors were looking for in their
simulation of the Müller–Lyer illusion: a change in
criterion, but no change in d 0 . Witt et al. (2015) Conclusion
criticized Rosenthal et al. (2009) for not comparing
these conditions in their SDT analysis and for instead In concluding, we wish to reiterate that the purpose
focusing on the change in d 0 between the no-beep and of this response is not to devalue the central point of
two-beep conditions. This analysis is implied to be Witt et al. (2015) concerning the misinterpretation of
erroneous as Witt et al. (2015) stated, ‘‘Theoretically, the signal detection theoretic criterion measure. There
the sound-induced flash illusion is an example of a is no doubt that this problem runs deep in the
perceptual bias, and therefore the illusion should literature, and their article is thus an important
present itself in the measure of c (or b), and not in d 0 ’’ warning to SDT users everywhere. Rather, our goal
(p. 3). If we take this statement at face value, are we to was simply to clarify specific supporting arguments
believe that the change in d 0 that occurs between the that we found to be inconsistent, or that we believed
unisensory and multisensory conditions in Rosenthal et were misrepresentative of previous work. The com-
al. (2009) is not a manifestation of the sound-induced mon thread in these arguments seems to be an
flash illusion? Clearly, this interpretation is unsatisfac- overreliance on the criterion measure in terms of
tory, as it is undeniable that the change in d 0 is at least explaining the perceptual processes associated with
somehow related to the multisensory conditions that each illusion, which may have arisen out of an
induce the illusion. eagerness on the part of the authors to provide

Downloaded From: https://jov.arvojournals.org/ on 08/14/2018

 Journal of Vision (2016) 16(11):18, 1–4 Knotts & Shams 4

Figure 3. Signal detection measures approximated for each of three different auditory conditions in Rosenthal et al. (2009, experiment
1, feedback condition). Both d 0 and c show significant change between each pair of conditions, except in the case of d’ between the
two multisensory conditions. Note that we have modeled shifts in the criterion as perceptual biases according to Witt et al. (2015).

support for the idea that the criterion measure can References
reflect perceptual bias in addition to response bias. We
hope that showing the danger in making these types of
oversimplifying assumptions provides a gentle warn- Morgan, M. J., Hole, G. J., & Glennerster, A. (1990).
ing of its own to readers and others using signal Biases and sensitivities in geometric illusions. Visual
detection theoretic models of perception going for- Research, 30, 1793–1810.
ward. Raslear, T. G. (1985). Perceptual bias and response
bias in temporal bisection. Perception & Psycho-
Keywords: signal detection theory, Müller-Lyer illu- physics, 38(3), 261–268.
sion, sound-induced flash illusion
Rosenthal, O., Shimojo, S., & Shams, L. (2009). Sound-
induced flash illusion is resistant to feedback
training. Brain Topography, 21(3–4), 185–192.
Acknowledgments Tudusciac, O., & Nieder, A. (2010). Comparison of
length judgments and the Müller–Lyer illusion in
monkeys and humans. Experimental Brain Re-
Commercial relationships: None. search, 207(3–4), 221–231.
Corresponding author: J. D. Knotts. Witt, J. K., Taylor, J. E. T., Sugovic, M., & Wixted, J.
Email: jknotts@g.ucla.edu. T. (2015). Signal detection measures cannot distin-
Address: University of California, Los Angeles, Los guish perceptual biases from response biases.
Angeles, CA, USA. Perception, 44(3), 289–300.

Downloaded From: https://jov.arvojournals.org/ on 08/14/2018