Sei sulla pagina 1di 13

Atten Percept Psychophys (2012) 74:1475–1487

DOI 10.3758/s13414-012-0330-z

An examination of the processing capacity of features


in the Thatcher illusion
Nick Donnelly & Katherine Cornes & Tamaryn Menneer

Published online: 16 June 2012


# Psychonomic Society, Inc. 2012

Abstract Detection of the Thatcher illusion (Thompson, redundancy gains resulted from supercapacity processing.
Perception, 9:483–484, 1980) is widely upheld as being We concluded that the oddity signalled by inversion of eyes
dependent on configural processing (e.g., Bartlett & Searcy, and mouths does not arise from positive interdependencies
Cognitive Psychology, 25:281–316, 1993; Boutsen, between these features.
Humphreys, Praamstra, & Warbrick, NeuroImage, 32:352–
367, 2006; Donnelly & Hadwin, Visual Cognition, 10:1001– Keywords Face perception . Response time models
1017, 2003; Leder & Bruce, Quarterly Journal of Experimental
Psychology, 53A:513–536, 2000; Lewis, Perception, 30:769–
774, 2001; Maurer, Grand, & Mondloch, Trends in Cognitive In the Thatcher illusion, faces appear grotesque as a
Sciences, 6:255–260, 2002; Stürzel & Spillmann, Perception, consequence of inverting the eyes and mouth in an
29:937–942, 2000). Given that supercapacity processing otherwise upright face (Bartlett & Searcy, 1993; Cornes,
accompanies configural processing (see Wenger & Donnelly, Godwin, & Wenger, 2011; Thompson, 1980).
Townsend, 2001), supercapacity processing should occur However, the grotesque appearance disappears when
in the processing of Thatcherised upright faces. The Thatcherised faces are inverted. Given that the Thatcher
purpose of this study was to test for evidence that the illusion is affected by inversion, the Thatcherisation of
grotesqueness of upright Thatcherised faces results from features is thought to affect relational (or configural)
supercapacity processing. Two tasks were employed: information within the faces (e.g., Boutsen, Humphreys,
categorisation of a single face as odd or normal, and a Praamstra, & Warbrick, 2006; Donnelly & Hadwin,
same/different task for sequentially presented faces. The 2003; Leder & Bruce, 2000; Lewis, 2001; Maurer,
stimuli were typical faces, partially Thatcherised faces Grand, & Mondloch, 2002; Stürzel & Spillmann, 2000).
(either eyes or mouth inverted) and fully Thatcherised Typically, the explanation of the Thatcher illusion is
faces. All of the faces were presented upright. The data from couched in terms of the dual-mode hypothesis (Bartlett &
both experiments were analysed using mean response times Searcy, 1993), which states that faces are processed in terms
and a number of capacity measures (capacity coefficient, the of features and of the relationships between features
Miller and Grice inequalities, and the proportional-hazards (configurations), but with the processing of configurations
ratio). The results of both experiments demonstrated some restricted to upright faces.
evidence of a redundancy gain for the redundant-target The aim of the present research was to test whether the
condition over the single-target condition, especially in grotesqueness of Thatcherised faces is subject to a form of
the response times in Experiment 1. However, there was configural or face superiority effect (Homa, Haver, &
very limited evidence, in either experiment, that the Schwartz, 1976; Pomerantz, Sager, & Stoever, 1977). In
relation to the Thatcher illusion, the issue of configural
superiority can be explored by measuring the processing
N. Donnelly (*) : K. Cornes : T. Menneer
capacity for detecting the oddity signalled by the inversion
School of Psychology, University of Southampton,
Southampton SO17 1BJ, UK of eyes and mouths. Processing capacity reflects the resources
e-mail: N.Donnelly@soton.ac.uk available to complete a task at a given time. It is calculated
1476 Atten Percept Psychophys (2012) 74:1475–1487

from the relationship between the amounts of work that can be Wenger, 2004) to determine the processing capacity for
achieved in the presence of different amounts of information Thatcherised faces.
(e.g., both eyes and mouth inverted vs. either eyes or mouth To the extent that the Thatcher illusion is processed
inverted). Importantly, measures of processing capacity configurally, then interaction or dependencies between the
provide an assessment of performance (work achieved) eyes and the mouth should allow more work to be achieved
that goes beyond that which can be inferred from simple in the full than in the partial Thatcher conditions. We therefore
differences in average response time (RT) data across hypothesised that we would find supercapacity processing for
conditions. Most importantly, processing capacity can be detecting oddity when both eyes and mouths are inverted
determined at all time points, rather than being based on relative to when only eyes or only mouths are inverted. Failure
a comparison of measures of central tendency. to find supercapacity would bring into question the presence
Three levels of processing capacity relate to the effect on of configural processing in generating the illusion. Moreover,
performance when variations in the amount of information given the status of the Thatcher illusion as the prime exemplar
(i.e., workload) occur (see Townsend & Nozawa, 1995). of the effect of configurality on face processing, any evidence
Limited capacity is signified by a decrement in performance that it does not result from configural processing would
as workload increases. Unlimited capacity occurs when require us to ask fundamental questions about configural
there is no change in performance as workload increases. face processing in general.
Finally, in supercapacity processing, positive interactions In the redundant-targets paradigm, performance is
between the channels cause performance to improve under compared in three conditions: A (here defined as inverted
increased workload (Wenger & Townsend, 2001). eyes), B (here defined as inverted mouths), and A plus B (here
Given that supercapacity is required for configural defined as inverted eyes and mouths). First, RTs are compared
processing (see Blaha, 2004, and Blaha & Townsend, in order to determine whether there is a gain for the redundant-
2006, for changes in capacity with configural learning; target condition as compared with the single-target conditions.
see Wenger & Townsend, 2001, for the relationship If such a redundancy gain is found, the critical issue is whether
between capacity and configural, or Gestalt, processing), this gain arises from supercapacity processing. Supercapacity
we hypothesised that supercapacity processing should be arises when the work done in the A-plus-B condition is greater
observed in detecting oddity in Thatcherised faces. With than the sum of the work done in the A and B conditions. The
respect to this hypothesis, it is important to be clear work done is calculated from the RT distributions, as
how configural processing is causal in the experiencing explained later. In examining this issue, we used a number
of the Thatcher illusion. There are two ways in which of measures: the proportional-hazards model (Allison, 2010;
the grotesqueness in the Thatcher illusion may be related to Cox, 1972), the capacity coefficient (Townsend & Nozawa,
configurality. First, inversion of the eyes and mouth might 1995), the Miller inequality (Miller, 1982), and the Grice
create highly atypical configurations, and such atypicality inequality (Grice, Canham, & Boroughs, 1984; see Townsend
causes the experience of grotesqueness. Second, the feature & Wenger, 2004, for a discussion of the relations among these
inversion of eyes and mouths may impede the ability to measures). All four measures were used to provide converging
encode the eyes and mouths within facial configurations that evidence for or against the hypothesis that supercapacity
are otherwise fluently processed. In this second hypothesis, processing of full Thatcherisation (eyes and mouth inverted)
Thatcherisation is associated with local difficulties in configural will occur relative to partial Thatcherisation (either the eyes or
processing. Nevertheless, and according to either version, the the mouth inverted). Average RTs alone cannot be used to
configurations computed for upright faces act as carriers for calculate capacity, and therefore cannot address the question
signals of oddity (grotesqueness). Evidence of superca- of configurality in the detection of Thatcherisation.
pacity in oddity detection would reveal a form of con-
figural superiority for the detection of oddity from
emergent features in Thatcher faces. Processing efficiency across time: methods of analysis
Here we manipulated the number of target (inverted)
features, to create partial Thatcher faces, in which either The formulae for calculating the capacity coefficient, the
eyes or mouths were inverted, as well as full-Thatcher faces, Miller inequality, and the Grice inequality are presented
in which both eyes and mouths were inverted. Changes in in the following sections, together with discussion of
the number of inverted features are known to correspond possible outcomes.
to perceived grotesqueness, with grotesqueness ratings
increasing as the number of inverted features increases Hazard ratios
from one to two (Cornes et al., 2011). By manipulating
the number of features inverted, we used the redundant- The hazard function provides a measure of the probability of
targets paradigm (e.g., Miller, 1982; Townsend & making a response at the next time point, given that the task
Atten Percept Psychophys (2012) 74:1475–1487 1477

is not yet completed (Wenger & Townsend, 2000). Formally, C(t) > 1, demonstrating that the work achieved in the full-
it is the probability that a response is made at a given time Thatcher condition is greater than the sum of the work
point (probability density function of task completion over achieved with the partial-Thatcher changes.
time) divided by the probability that a response has not yet
been made up to and including a given time point (the survivor The Miller inequality (Ingvalson & Wenger, 2005; Miller,
function1). The hazard function can be used as a measure of 1982; Townsend & Nozawa, 1995)
the amount of work that can be achieved in a given unit of
time, and as such lends itself to measuring the processing SF ðtÞ  SP1 ðtÞ  SP2 ðtÞ þ 1  0 ð2Þ
capacity of a system—that is, the relationship between
amounts of work achieved under different conditions
The Miller inequality is given in Eq. 2, where t is a given
(Townsend & Ashby, 1978; Wenger & Gibson, 2004).
time point, SF(t) is the survivor function for the full-
The proportional-hazards model (see Allison, 2010; Cox,
Thatcher condition, SP1(t) is the survivor function for the
1972; Wenger & Gibson, 2004) can be used to determine the
first partial-Thatcher condition, and SP2(t) is the survivor
ordering of hazard functions, and as such allows a test of
function for the second partial-Thatcher condition.
whether work is achieved earlier in the full-Thatcher condition
A violation of the Miller (1982) inequality implies super-
than in the partial-Thatcher conditions by calculating the ratio
capacity. If SF(t) 0 SP1(t) + SP2(t), it can be clearly seen that
of the hazard function for the full-Thatcher condition to that
the Miller inequality will not be violated (the left-hand side
for the partial-Thatcher conditions. Because this measure
[LHS] will be 1). Similarly, the inequality will not be
relies on the relative ordering of the hazard functions, rather
violated when SF(t) > SP1(t) + SP2(t) (LHS will be
than on precise values, it is more robust than alternative
greater than 1). A violation of the Miller inequality will
methods when there are relatively few data points (Allison,
occur when SF(t) + 1 < SP1(t) + SP2(t). That is, at time
2010; Wenger & Gibson, 2004).
t, the probability that a response has not occurred for the full-
An alternative method for measuring capacity using the
Thatcher condition is lower than the summed probabilities for
hazard function is the capacity coefficient (Townsend &
the partial-Thatcher conditions, and this difference at least 1.2
Nozawa, 1995). This measure is calculated using Eq. 1,
In other words, responses occur faster for the full-Thatcher
where t is a given time point, HF(t) is the integrated hazard
condition than they do for either of the partial-Thatcher
function for the full-Thatcher condition, HP1(t) is the
conditions. Given that the Miller inequality is a more
integrated hazard function for one partial-Thatcher condition,
conservative test of supercapacity than the capacity coefficient,
and HP2(t) is the integrated hazard function for the other
a violation is evidence for extreme supercapacity (see Ingvalson
partial-Thatcher condition.
& Wenger, 2005, p. 19; Townsend & Wenger, 2004).
HF ðtÞ
CðtÞ ¼ ð1Þ
HP1 ðtÞ þ HP2 ðtÞ The Grice inequality (Grice et al., 1984; Ingvalson
& Wenger, 2005; Townsend & Nozawa, 1995)
Given that the hazard function provides a measure of the
amount of work done in a given unit of time, its integral The Grice inequality (Eq. 3) is computed by subtracting the
provides a cumulative measure of work done (Townsend & survivor function for the full-Thatcher condition from the
Ashby, 1978; see also Wenger & Townsend, 2000). This survivor function of either the eye condition or the mouth
cumulative hazard function is calculated as the negative log condition, depending on which of these values is smaller:
of the survivor function. When the work done at a given
min½SP1 ðtÞ; SP2 ðtÞ  SF ðtÞ  0: ð3Þ
time point is less for the full-Thatcher [HF(t)] than for the
sum of both partial-Thatcher [HP1(t) + HP2(t)] conditions, As well as measuring processing capacity, the Grice
capacity is limited, in that only a certain amount of inequality can also be used to determine a redundancy gain
information can be processed, and any extra information in a finer-grained way than can be achieved with mean RTs.
does not cause more work to be achieved. Limited A violation of the Grice inequality implies limited capacity.
capacity therefore occurs when HF(t) < HP1(t) + HP2(t), such If SF(t) > SP1(t) or SF(t) > SP2(t), then the Grice inequality is
that C(t) < 1. Unlimited capacity is defined as when violated (LHS < 0). In other words, a violation occurs at time t
HF(t) 0 HP1(t) + HP2(t), such that C(t) 0 1. Super- if the probability of not having responded is lower in one of
capacity is characterised by HF(t) > HP1(t) + HP2(t), such that the partial-Thatcher conditions than in the full-Thatcher

1
Survivor function 0 1 – cumulative distribution function, which 2
Given that the maximum for any probability is 1, the requirement that
provides the probability of no response having occurred as a function SF(t) is at least 1 less than SP1(t) + SP2(t) makes the Miller (1982)
of time. inequality a conservative test of supercapacity.
1478 Atten Percept Psychophys (2012) 74:1475–1487

condition. Put simply, a violation occurs when responses are (full-Thatcher faces) (see Fig. 1).3 In total, there were 64
faster in at least one of partial-Thatcher conditions than in the face stimuli. The faces were manipulated in Adobe Photoshop
full-Thatcher condition. A violation therefore provides with the blur tool used to remove high-contrast edges caused
evidence for limited capacity, in which there is no capacity by manipulating the images. Whole images were blurred
for processing the extra information in the full-Thatcher using Gaussian blur with a radius of one pixel. The stimuli
condition, and responses are actually slower than the faster of subtended 6.7° × 8.3° of visual angle, with the face occupying
the two partial-Thatcher conditions. The requirement for 4.1° × 6.3° within the image, and, perhaps most critically for
slower responses in the full-Thatcher condition causes the task, the distance between the top of the eyebrows and the
a violation of this inequality to be evidence for extreme bottom of the mouth was 2.9° (see Fig. 1).
limited capacity (see also note 8 of Ingvalson & Wenger, In order to address the hypothesis that detection of the
2005; Townsend & Wenger, 2004): A full-Thatcher condition Thatcherised features in upright faces would be marked by
is actually less helpful than at least one of the partial-Thatcher supercapacity processing, only upright faces were presented.
conditions.
Unlike the capacity coefficient and the Miller inequality, Procedure Participants were presented with a single, typical
the Grice inequality is not reliant on a comparison between partial- or full-Thatcher face. The participants were asked to
the full-Thatcher and both partial-Thatcher conditions. decide whether the face was “odd” or “normal” as quickly and
Rather, it compares the full-Thatcher condition with the accurately as possible. The stimulus was displayed until re-
faster of the partial-Thatcher conditions, regardless of sponse. Given the nature of the response (“odd” or “normal”),
whether one partial-Thatcher condition was relatively fast the task was to detect oddity, or grotesqueness, within the
and the other relatively slow. stimulus.
The experiment comprised 768 trials, divided into four
blocks. Each Thatcherised face was presented twice in each
Experiment 1 block, to give 32 eye, 32 mouth, and 32 eye-and-mouth
trials. Each unchanged face was presented six times in each
The aim of Experiment 1 was to test whether inverted block to give 96 typical trials. In total, each participant
mouths and eyes, presented in an upright face context, viewed 128 eye, 128 mouth, and 128 eye-and-mouth
produce supercapacity processing, or whether responses to Thatcher trials, and 384 typical trials. The participants had
full-Thatcher faces are equivalent to the combination of a short, self-paced break between blocks. Face identity and
responses to inverted eyes and inverted mouths in partial- type of manipulation were randomised within all blocks.
Thatcher faces. Inverted faces were not tested, as we did not The primary aim in deciding the frequency of each trial
hypothesise supercapacity processing for inverted faces. type was to equate the number of “odd” and “normal”
responses, such that the probability of finding an inversion
Method in any given face was .5. However, the trial proportions
chosen may increase the chance of finding a redundancy
Participants A group of 12 undergraduate students partici- gain by artificially inhibiting responses to one of the partial-
pated in Experiment 1 for course credit. The participants had Thatcher faces. Such inhibition can occur for two reasons.
a mean age of 20.5 years (SD 0 3.6). Two of the participants Firstly, given no inversion for one feature (e.g., the eyes),
were male, ten were right-handed, and all had self-reported the probability of an inversion for the other feature is .25
normal or corrected-to-normal vision. (e.g., 32 mouth-inversion trials vs. 96 typical trials). This
probability is lower than the overall probability of an inversion
Apparatus The experiment was conducted using custom E- (.5), and will therefore inhibit responses to single-inversion
Prime software (Psychology Software Tools, Inc., Sharpsburg, trials (e.g., responses to mouth-inversion trials will be slower
PA), with stimuli displayed on a 15-in. monitor set to a refresh than those to full-Thatcher trials) (Mordkoff & Egeth, 1993;
rate of 60 Hz and a resolution of 1,280 × 1,024. Responses
3
were made using the left and right mouse buttons, with the A control experiment was run with the eyes and mouths outside of
buttons counterbalanced across participants. The participants their facial contexts. Using a same/different paradigm, participants
were presented with pairs of identical eyes and mouths of similar
were seated approximately 57 cm from the screen.
orientations (either both upright or both inverted) or different orienta-
tions (one upright and one inverted). There was no difference between
Stimuli Sixteen greyscale female faces were obtained from responses to the eyes and mouths in terms of RT or accuracy, suggest-
the Stirling Picture Database. These 16 faces were used to ing that the eyes and mouths were matched for overall pair-wise
difficulty of discrimination. Equal discriminability reduces the poten-
create four sets of stimuli: no stimulus manipulation (typical
tial for bias towards one feature over the other, and is therefore less
faces), only the eyes or only the mouths inverted (partial- likely to encourage possible frequency effects such as the ones de-
Thatcher faces), and both the eyes and mouths inverted scribed in the Procedure.
Atten Percept Psychophys (2012) 74:1475–1487 1479

conducted both for individual participants and over all


participants. Secondly, given a redundancy gain, the
measures of processing efficiency were calculated in
order to establish the capacity levels underlying the
redundancy gain. All of the analyses were conducted
only on Thatcher trials, in which a feature inversion had
occurred, in order to compare between the partial- and full-
Thatcher conditions. All post-hoc t tests were Bonferroni
corrected by adjusting the p value.

Redundancy gain: analyses of group mean and individual


RTs To test for an overall redundancy gain, the effect of the
number of feature changes (partial vs. full Thatcher) on RTs
and error rates was first examined at the group level. In
order to minimise positive skew in the RT data, and hence to
eliminate the need for outlier trimming, reciprocal RTs were
analysed (Ratcliff, 1993) using the mean reciprocal RT over
correct trials. Nontransformed mean RTs were also analysed
(see note 5 for the results). The error data were transformed
pffiffiffi
using the arcsine transformation, arcsin ð xÞ, where x is the
percentage error. This transformation was required to
normalise the binomial distribution of the accuracy data.
One-way repeated measures ANOVAs revealed a significant
Fig. 1 Example stimuli from Experiments 1 and 2, with a typical face effect of feature change for RTs and error rates, Fs(2, 22) 0
(top left), eye partial-Thatcher (top right), mouth partial-Thatcher (bot- 22.03 and 35.97, both ps < .01, partial eta-squareds (ηp2s) 0 .67
tom left) and full-Thatcher (bottom right). Note the example of the face
and .77, respectively (see Fig. 2). Follow-up t tests revealed a
image is not one of the stimuli used in the experiment, but an illustra-
tive example of the Thatcher illusion as instantiated in this study redundancy gain, with faster responses and higher accuracy in
the full-Thatcher condition than in both partial-Thatcher
Mordkoff & Yantis, 1991). It is worth noting that the same conditions [all ts(11) > 6.36, p < .01]. There was no significant
contingencies are not implied by the presence of an inversion: difference between the eye and mouth conditions for RTs or
Inversion of one feature (e.g., the eyes) is not predictive of the errors [ts(11) 0 2.26 and 2.68, ps 0 .14 and .07, respectively].4
inversion of the other feature (e.g., the mouth), because given As the capacity measures will be examined at the level of
an eye inversion, the proportion of trials with a mouth the individual, each participant’s RT data were also analysed
inversion is .5. Secondly, the RTs for the single-inversion trials separately to test for a redundancy gain for each individual.
can be artificially slowed if the participant preferentially For each participant, a separate one-way independent-
attends to one location or another (Mullin, Egeth, & samples ANOVA5 was conducted for the Feature Change
Mordkoff, 1988). For example, if the participant attends to factor, with three levels (eye and mouth vs. eye vs. mouth)
the eye location, an eye inversion would result in faster and with up to 128 trials in each condition. The effect of
responses than would a mouth inversion. This difference feature change was significant for ten of the 12 participants
would not affect the responses to full-Thatcher trials but [F(2, 325) to F(2, 371) > 3.5, ps < .04, ηp2s 0 .02 to .2],
would slow the mouth-inversion trials relative to the other approached significance for one participant [F(2, 322) 0
trial types, causing an overall slowing for single-inversion 2.60, p 0 .08, ηp2 0 .02], and was not significant for the last
trials. The participants were not informed about these participant [F(2, 302) 0 1.32, p 0 .27, ηp2 0 .01]. Follow-up t
probabilities of co-occurrence. tests were conducted on the ten participants showing a signif-
icant effect of feature change. For five of the participants, RTs

Results 4
Nontransformed RTs were also analysed, at both the individual and
group levels. The results revealed a pattern similar to that observed
Before assessing processing efficiency, analyses were with reciprocal RTs, except that one fewer participant exhibited a
conducted on the performance data (RT and accuracy) in order redundancy gain at the level of the individual.
5
to establish whether a redundancy gain occurred for detecting RTs across the three conditions are from the same participant, but a
within-participants ANOVA could not be employed here, because it
grotesqueness in the full-Thatcher condition as compared with was not possible to pair the RTs from each condition with those from
the partial-Thatcher conditions. These analyses were the other conditions.
1480 Atten Percept Psychophys (2012) 74:1475–1487

LIFETEST procedure, with the life-table method and a width


(bin size) of 25 ms (see Allison, 2010).6
For the group of participants exhibiting a redundancy
gain (n 0 5/12), overall capacity measures at the level of
the group (Fig. 3) showed supercapacity at early time points
(capacity coefficient > 1, violation of the Miller inequality),
as well as a little evidence for extreme limited capacity at
late time points (violation of the Grice inequality). For the
group of participants not exhibiting a redundancy gain
(n 0 7/12), there was no evidence of supercapacity (capacity
coefficient < 1, one very minor violation of the Miller
inequality) or of extreme limited capacity (violations of the
Grice inequality). In sum, there was only modest evi-
dence for supercapacity processing in five out of 12
participants, and for most of these the evidence was
limited to very early time points.
For completeness, at the level of individuals, there was
very little evidence of extreme supercapacity from the Miller
inequality (few violations, all values close to or above zero),
with the few violations that did occur occurring at early time
Fig. 2 Mean response times (RTs, in milliseconds) and error rates points ( < 800 ms). Those participants demonstrating the
(percentages) from Experiment 1 (black) and Experiment 2 (grey).
Error bars represent one standard error
strongest violations were those that exhibited a redundancy
gain (black lines, Fig. 4). There was some evidence of
extreme limited capacity from the Grice inequality for some
were faster in the full-Thatcher condition than in one of the participants, with violations at some time points (values
partial-Thatcher conditions (all ps < .05). For the remain- below zero). The participants demonstrating violations were
ing five participants, RTs were faster in the full-Thatcher ones who did not exhibit a redundancy gain (grey lines,
condition than in both of the partial-Thatcher conditions (all Fig. 4). From the capacity coefficients, there was some
ps < .05), demonstrating a redundancy gain. evidence of supercapacity, with values exceeding 1 for the
majority of time points for one participant, and at some early
Processing efficiency across time The RT analyses revealed time points ( < 750 ms) for five participants. All but one of
a significant redundancy gain at the level of the group, and these participants exhibited a redundancy gain (black lines,
at the level of the individual for five participants. We wished Fig. 4). For mid-to-late time points, the capacity coefficients
to determine whether or not this redundancy gain arose due were between 0 and 1, indicating limited to unlimited
to supercapacity processing. Mean-RT analyses alone cannot capacity, but no evidence of supercapacity processing.
be used to answer questions regarding processing capacity,
particularly when capacity changes over time, and there- Discussion
fore the measures of processing capacity defined earlier
were employed. The analysis revealed a redundancy gain at the level of the
The three panels in Fig. 3 show the mean values for the mean, as well as redundancy gains at the level of the individual
capacity coefficients, Miller inequalities, and Grice inequalities for five of the 12 participants. For most participants, the capacity
calculated across groups (those exhibiting a redundancy gain data are inconsistent with interdependencies between features
vs. those not), and Fig. 4 presents these measures for all 12
participants individually. Given that the three methods are
6
nonparametric, no transformation of RTs was required before Note that calculation of the capacity coefficient and of the Miller
analysis. RTs for incorrect trials were removed before analysis, inequality only makes sense when all three conditions (full Thatcher,
eye inversion, and mouth inversion) have a survivor function value,
and for each incorrect-trial RT, the most similar correct-trial RT
and the capacity coefficient cannot be calculated when the hazard
was also removed from the same condition. This removal of function is zero for both of the partial-Thatcher conditions. On the
twin RTs is standard practice and is necessary because there other hand, the Grice inequality requires that the full-Thatcher condi-
were two possible responses (odd and normal), so for every tion and only one of the single-inversion conditions have survivor
function values. Given that the number of time points that have
incorrect trial that could be considered a guess, there was also a
survivor function values will vary across condition, the number of time
correct trial that could be considered a guess (Eriksen, 1988). points that can be calculated across capacity and the Miller and Grice
The survivor functions were calculated using the SAS inequalities will also differ.
Atten Percept Psychophys (2012) 74:1475–1487 1481

Capacity Coefficient

Capacity Coefficient
1.5
5
4
1.0
3
2
0.5 1
0
0 1000 2000 3000 0 1000 2000 3000 4000
Time from earliest response (ms) RT bins

1.0 1.0

Miller Inequality
Miller Inequality

0.5
0.5
0.0

0.0 0 1000 2000 3000 4000


RT bins
0 1000 2000 3000
RT bins
1.0
Grice Inequality
0.2
Grice Inequality

0.5
0.1
0.0
0.0
0 1000 2000 3000 4000
RT bins
0 1000 2000 3000 4000 Fig. 4 Processing efficiency data from Experiment 1. Each line rep-
RT bins resents the efficiency measure for one participant over time. Black lines
represent values from those participants who exhibited a redundancy
Fig. 3 Processing efficiency data from Experiment 1, presented at the gain, and grey lines represent values from those who did not
level of the group. Black lines represent the mean values for the group
of participants exhibiting a redundancy gain, and grey lines represent
the mean values for the group who did not. The capacity coefficient
means were calculated by lining up the earliest response of each gain arises from supercapacity, and no evidence that
participant before averaging across relative time bins supercapacity is necessary for a redundancy gain, especially
at mid-to-late time points. No extra work is necessarily
achieved in the full-Thatcher condition, as compared with
that serve to boost performance. Therefore, the results demon- the total work across both partial-Thatcher conditions. In
strate that supercapacity is not necessary, nor automatic, for addition, there is little evidence that the redundancy gain,
detection of the Thatcher illusion, and some participants were when present, emerges because of positive interactions
able to complete the task with limited capacity. However, there between features. These results could be explained by a race to
were some participants, particularly those demonstrating a re- completion of two independently processed features.
dundancy gain, for whom there was some evidence of process- In these tasks, our reasoning was that participants sought
ing reaching supercapacity, with nonadditive processing of eyes evidence of inversion (and, hence, oddity). The presence of
and mouths. one inverted feature would require an “odd” response, while
The results suggest that the discrimination between the presence of a typical feature would not necessarily
Thatcher and typical faces does not rely on supercapacity indicate a “normal” response. The gain in RTs for the
processing and can occur using a limited to an unlimited redundant inverted feature (full over partial Thatcher) does
capacity system. There is little evidence that the redundancy indeed suggest that participants were using the inverted
1482 Atten Percept Psychophys (2012) 74:1475–1487

feature(s) to conduct the task. However, we should note that would allow for self-termination after examination of a single
there is nothing to prevent participants from assigning the feature, rather than exhaustive processing of both features.
same task relevance to upright as to inverted features. In this Such a change would suggest that configural processing was
case, the presence of both types of information in the partial- no longer taking place (Wenger & Townsend, 2001) and
Thatcher condition would engage additional resources for would therefore affect the inferences regarding capacity made
processing relative to the full-Thatcher condition. The effect in Experiment 1. Thirdly, Experiment 1 might not have
of this would be to suppress the measure of processing provided strong evidence for supercapacity if participants
capacity. Although this is possible, the account is highly used typicality as well as oddity to conduct the task, as we
unlikely to hold, as it would require participants to engage discussed earlier.
in the simultaneous detection of two types of targets. In In Experiment 2, we explored whether the failure to
other types of visual-processing tasks, it is known that observe supercapacity in Experiment 1 might have resulted
performance is compromised when participants are from the reasons described above (categorisation task
required to, for example, search for and detect two different demands, eye movements, and two sources of information
targets simultaneously (Menneer, Barrett, Phillips, Donnelly, in the partial-Thatcher condition). Using the same stimulus
& Cave, 2007). set, we adopted the procedure employed by Ingvalson and
It might be possible that participants could detect a feature Wenger (2005), in which same/different judgements were to
inversion but not experience grotesqueness. If so, the results be made to successively presented face pairs. The design
would not require reference to grotesqueness at all. This is, allowed for “same” and “different” responses to be orthogonal
however, highly unlikely to be the case. Firstly, the to the presentation of typical and Thatcherised faces. In doing
phenomenon of grotesqueness in the illusion is experienced so, Experiment 2 would prevent the use of a single absolute
automatically (e.g., Bartlett & Searcy, 1993), with a value (odd/normal) to complete the task. While this task could
correlation between the timings of activation of the still be terminated on the first feature of the second face,
fusiform face area and the amygdala when viewing because only one difference was required in order to
Thatcherised faces (Donnelly et al., 2011; Rotshtein, provide a “different” response, the probe face was pre-
Malach, Hadar, Graif, & Hendler, 2001), and secondly, sented sufficiently briefly that eye movements could not
the magnitude of perceived grotesqueness is higher for full- be made in response to these faces. Nevertheless, eye
than for partial-Thatcherised faces (Cornes et al., 2011). movements were measured, and data were excluded
Therefore, it is highly unlikely that any strategy for perceiving from trials on which eye movements were made.
inversion without the experience of grotesqueness could
succeed. In addition, our analyses concerned comparisons
across full-Thatcher and partial-Thatcher conditions. Method
The presentation of features remains constant across
these conditions, and therefore any particular instance Participants A group of 15 student volunteers were recruited
of a feature in which grotesqueness was not apparent as participants. Three of the participants were excluded, as their
would be the same across both partial- and full-Thatcher average error rates across conditions exceeded 25 %.7 The
faces. mean age of the remaining participants was 25.8 years (SD 0
6.6). Three of the participants were male, 11 were right-handed,
and all had self-reported normal or corrected-to-normal vision.
Experiment 2
Apparatus The stimuli were presented on a 21-in. monitor
There are at least three possible reasons why a redundancy with a refresh rate of 60 Hz and a resolution of 1,280 × 1,024.
gain could be found in Experiment 1 without strong Responses were made via a two-button response box, with the
evidence for supercapacity. The first is that the task of buttons counterbalanced across participants. Eye movements
categorising faces as “odd” or “normal” could have were recorded using an SR Research EyeLink 1000 eyetracker
encouraged participants to focus on the indicative features and the associated software. Although stimulus viewing was
(eyes or mouth) to make the categorisation. Secondly, the binocular, recordings of eye movements were made from the
2.9° distance between eyebrows and mouth may have right eye only. A chin rest and a forehead rest were used, with
encouraged eye movements, which in turn are likely to the participants seated 57 cm from the monitor.
encourage the selective processing of individual features more
than when eye movements are prohibited, particularly because 7
An error rate exceeding 25 % may seem high, but the faces were
eye movements are frequently made to eyes and mouths in the
small (1º between eyes and mouth; see the Stimuli section), the
processing of faces (Smith, Gosselin, & Schyns, 2004; Spezio, changes were minimal (see Fig. 1), and the presentations were brief
Adolphs, Hurley, & Piven, 2007). Such serial processing (150 ms for second face, as described in the Procedure section).
Atten Percept Psychophys (2012) 74:1475–1487 1483

Stimuli The stimuli were the same as in Experiment 1, the practice trials were not recorded. Participants were asked
except for the following. Only ten of the 16 faces were used, to keep their eyes still when viewing the stimulus. If more
in order to reduce the number of trials and to prevent the than one fixation was made on the second face, the trial and
participants from becoming uncomfortable during the data were discarded, which resulted in two trials being
eyetracking process. The ten faces used were randomly discarded from the analysis across all participants.
selected from the 16. Secondly, the faces were presented
with a smaller visual angle of 2.1° × 2.6°, giving a Results
maximum of 1.0° vertical distance from the top of the
eyebrows to the bottom of the mouth. A smaller stimulus was The data were analysed as in Experiment 1 unless otherwise
used in order to allow processing of the face within a single stated. All analyses were conducted only on “different”
fixation, and thus to minimise serial or piecemeal processing trials that contained a typical face, such that a double feature
of the eyes and mouth. change (typical paired with full Thatcher) exhibited a
smaller change in grotesqueness than did a single feature
Procedure There were four variants of each face: one typical, change (typical paired with partial Thatcher). A double
two partial-Thatcher faces, and one full Thatcher. Each variant feature change also occurred between a Thatcherised eye
of a face was paired with the three other variants of the same face and a Thatcherised mouth face, but the difference
face, giving six pairs. Two randomly selected faces were would not represent a change in the total amount of
reserved in order to provide 16 practice trials. For the eight grotesqueness. For completeness, the data for all types
faces used in the experimental trials, this process created 48 of trial are presented in Fig. 5.
pairs of faces matched for identity but varying in structure. Trials containing a typical face (i.e., those analysed)
The order of presentation of the two faces in each pair was could be presented with the typical face appearing first or
counterbalanced, providing 96 “different” trials, of which 32 second. Bonferroni-corrected t tests on reciprocal RTs and
were eye change only, 32 were mouth change only (either arcsine error rates revealed shorter RTs for both single-
typical to partial Thatcher or partial Thatcher to full Thatcher), change conditions, as well as lower error rates for the mouth
and 32 were double change (typical to full-Thatcher faces or change condition when the typical face was presented first
partial-Thatcher eye change to partial-Thatcher mouth than when it was presented second (all ts > 2.97, ps < .05).
change). Each variant was also matched with itself to create There was no effect of face order for the double-change
32 “same” pairs, which were each presented three times condition or for the eye change condition on error rates
throughout the experiment, to provide 96 “same” trials. There
were a total of 192 experimental trials. The trial frequencies
were the same as had been presented in Experiment 1, in that
the numbers of “same” and “different” responses were
equated, at the expense of artificially increasing the chance of
finding a redundancy gain (see the Exp. 1 Procedure).
The eyetracker was calibrated within 0.3° accuracy for
each participant using a nine-point calibration display. Each
trial began with a drift correction dot, which was displayed
until the participant fixated it within 0.5°. Calibration was
repeated if this level of accuracy was not achieved. After the
drift correction, the first face was displayed for 500 ms,
followed by a blank screen for 500 ms, then the second face
for 150 ms. This timing was chosen in response to piloting,
with the aim of avoiding multiple fixations on the second
face. During practice trials, the second face was presented for
250 ms to allow the participants to familiarise themselves with
the task. In all trials, a response screen of four centrally located
question marks was shown until response.
The participants were instructed to decide whether pairs
of sequentially presented faces were the same or different.
This task therefore differed from that in Experiment 1, in
that responses could not depend on the mere detection of Fig. 5 Mean response times (RTs, in millseconds) and error rates
grotesqueness. The participants were told to respond as (percentages) from Experiment 2 for each type of trial. Error bars
quickly as possible while minimising errors. The data from represent one standard error
1484 Atten Percept Psychophys (2012) 74:1475–1487

3.0

Proportional hazards ratio


(all ts < 1.88, ps > .26). This analysis shows that there are
some differences in the levels of difficulty within each type 2.5
of single-change trial. However, such differences were to be
expected, and given that both orders were used for each pair 2.0
of faces, any effects of order would cancel out throughout 1.5
the experiment.
1.0
Redundancy gain: analyses of group mean and individual 0.5
RTs One-way ANOVAs across all participants revealed a
0.0
significant main effect of feature change for errors, F(2, 22)

Overall
Individual participants
0 13.24, p < .01, ηp2 0 .55, but no effect for RTs, F < 1, ηp2 0
.05 (see Fig. 5). Error rates were lower in the double-change
condition than in both single-change conditions (both ts >
4.28, ps < .01), and there was no difference between the 1.0

Miller Inequality
single-change conditions (t < 1).
One-way ANOVAs for individual participants were used to
0.5
analyse the reciprocal RTs, with up to 16 trials in each
condition. One of the 12 individual-participant ANOVAs
0.0
gave a significant effect of feature change, F(2, 47) 0
5.75, p < .01, ηp2 0 .20, and one showed a trend
towards a significant effect, F(2, 43) 0 3.07, p 0 .06, -0.5
ηp2 0 .13. All other ANOVAs showed no significant 0 1000 2000
effect (all Fs < 1.79, ps > .18, ηp2s 0 .02 to .08). For RT bins
the participant exhibiting an effect of feature change,
RTs were faster in the double-change (full-Thatcher) 1.0
Grice Inequality

condition than in the eye change (partial-Thatcher) con-


dition, t(30) 0 3.94, p < .01, but not than in the mouth 0.5
change (partial-Thatcher) condition, t < 1. These results
do not exhibit a redundancy gain at the level of the 0.0
individual.
There was no evidence for a redundancy gain in the RT
-0.5
data at the level of the mean, as well as at the level of
individuals.8 However, the error data did suggest a redundancy 0 1000 2000 3000
gain, implying a speed–accuracy trade-off in which RTs were RT bins
speeded in the single-change conditions at the expense of Fig. 6 Processing efficiency data from Experiment 2. For the hazards
accuracy. If accuracy levels were equal across the single- ratios, the first 12 bars represent the individual participants, and the
change and double-change conditions, a redundancy gain in 13th bar represents the ratio calculated over all participants. For the
RTs may have been found, because single-change RTs may Miller and Grice measures, the black lines represent the mean values
over all participants, and each grey line represents the efficiency
have needed to be longer to achieve the levels of accuracy measure for one participant over time
exhibited in the double-change condition. Therefore, the
processing capacity measures will be calculated. Given that
the capacity measures were calculated after incorrect and PHREG procedure in SAS. Figure 6 shows the Cox
twinned correct RTs were removed, the RTs analysed proportional-hazards ratios, Miller inequalities, and Grice
represent correct nonguessed responses, and therefore inequalities at the levels of the group and the individual.
compensate for a speed–accuracy trade-off. Schoenfeld residuals were calculated for the RTs over all
participants and change conditions (eye, mouth, or both),
Processing efficiency across time The data were analysed as using the RESSCH procedure in SAS. There was no signifi-
for Experiment 1, except that the Cox proportional-hazards cant correlation between RTs and the Schoenfeld residuals,
ratio was calculated instead of the capacity coefficient confirming that the assumption of constant proportionality
because of the reduced number of trials, using the was met, and therefore the RTs did not require trimming
before analysis.
8
As in Experiment 1, nontransformed RTs were also analysed, revealing At the level of the group, the proportional-hazards ratio
the same pattern of results observed for reciprocal RTs. was calculated over all participants, with stratification by
Atten Percept Psychophys (2012) 74:1475–1487 1485

participants to control for heterogeneity between the partici- The lack of a redundancy gain for grotesqueness
pants. The ratio was not significantly different from 1 (p 0 detection here is in contrast with those of Bradshaw
.125), indicating that the number of changes had no effect on and Wallace (1971; also replicated by Perry, 2008, and
processing capacity. To examine changes in capacity over time, Perry, Blaha, & Townsend, 2008), who found that the
the Miller and Grice inequalities were calculated and showed speed of same/different face judgements increased with the
no evidence of supercapacity at any time point (no violations of number of differences between the faces. The difference
the Miller inequality), and some evidence for limited capacity between the results could be explained by the presentation
(violations of the Grice inequality). of the stimuli, which were presented simultaneously by
At the level of individuals, three participants had hazards Bradshaw and Wallace, but successively here. The simultaneous
ratios that were significantly greater than 1 (all ps ≤ .05), presentation used in previous research may have encouraged
indicating an increase in processing capacity between one piecemeal comparison between the faces, and thereby increased
and two changes (eyes or mouth vs. eyes and mouth). For the probability of attending to a changed feature early in the trial
the remaining nine participants, and over all participants when there were two changes as compared with when there was
together, the hazards ratios were not significantly different only one. In addition, despite being face-like, the faces used by
from 1 (all ps > .2). For the changes in capacity over time Bradshaw and Wallace were sketch-like Identi-Kit faces, which
(Miller and Grice inequalities), one participant, who also may have discouraged configural processing (Leder, 1996).
had a hazards ratio greater than 1, exhibited a large violation
of the Miller inequality, demonstrating supercapacity at the
early time points ( < 625 ms). The same participant showed General discussion
no violations of the Grice inequality, indicating that capacity
was not limited. Two other participants had a few minor Across successive same/different and categorisation tasks,
violations of the Grice inequality at early time points. whether or not eye movements were allowed, similar results
However, the remaining participants had larger violations of held: The distinguishing of Thatcherised faces from typical
the Grice inequality, implying limited capacity, particularly at faces generally occurred with capacity processing that was
early time points ( < 850 ms). sub-supercapacity (i.e., was limited to unlimited), such that
the grotesqueness signals in full-Thatcher faces rarely go
Discussion beyond the sum of grotesqueness signalled by partial-
Thatcher faces. This finding is contrary to what was
The lack of evidence for supercapacity in Experiment 1 expected. Furthermore, although a simple redundancy gain
might have been due to the task characteristics allowing was reliably found in Experiment 1, where a categorisation
piecemeal sampling of the facial features. In Experiment 2, response was required, it was not found in Experiment 2,
such sampling was prevented by restricting eye movements. where a same/different response was required. This result
However, there was no redundancy gain in RTs for any was obtained despite the fact that the experiments were
individual or at the level of the mean. The lack of a conducted with trial frequencies that should have biased
redundancy gain suggests that there was no benefit of towards finding a redundancy gain. We therefore conclude
two changes (typical as compared with full Thatcher) that the signalling of grotesqueness in Thatcher faces can be
over the single-change conditions (typical as compared aided when signalled by two features as compared with one,
with partial Thatchers). In addition, the capacity analyses but this facilitation only occurs when categorisation
provided very little evidence for the supercapacity processing responses to single faces are required. Even then, the
of feature inversion in upright faces. The results are consistent processing capacity data are consistent with simple,
with inverted eyes and mouths being processed independently parallel horse-race processing, and possibly with serial
in upright faces, or as having interdependences that are not processing on occasion (when extreme limited capacity
excitatory, thus failing to produce supercapacity. is found). These data are consistent with two accounts.
Despite the difference in the tasks, and the measures First, no interactions formed across inverted eyes and
taken to encourage configural processing, these results agree mouths serve to speed the detection of grotesqueness.
with those from Experiment 1 by demonstrating that Second, inhibitory interactions exist between feature-
processing of grotesqueness in the Thatcher illusion is processing channels. In the absence of other data,
not at supercapacity when participants can terminate on Occam’s razor suggests that we should go with the most
the first feature completed. There is, however, evidence parsimonious account and thus conclude that there is
that such experimental designs sometimes encourage little evidence of interactions across eyes and mouths
analytic, nonconfigural perception of features in faces (Wenger in the Thatcher illusion.
& Townsend, 2001), while exhaustive processing promotes While the stimuli and tasks were not equated across
findings of supercapacity (Wenger & Townsend, 2006). Experiments 1 and 2, a comparison across the two does
1486 Atten Percept Psychophys (2012) 74:1475–1487

reveal something of interest: While responses were faster to on the Thatcher illusion could be accounted for by shifts in
eyes than to mouths in Experiment 1, there was a trend in perceptual sensitivity and decision criteria (in terms of GRT, a
the opposite direction in Experiment 2. The shifting trend is violation of perceptual and decisional separability) across
also reflected in the accuracy data, with a significant orientations. Their data suggested that within-stimulus
interaction across experiments for accuracy [F(2, 44) 0 interactions between internal and external features in
3.19, p 0 .05], although it misses significance for RTs Thatcher faces were limited. We added to those data here by
[F(2, 44) 0 2.36, p 0 .11]. The two tasks were different, reporting very limited evidence of oddity detection occurring
and therefore there was no a priori expectation that RTs in the with supercapacity.
full-Thatcher condition would be matched across experi- The present findings provide some initial evidence about
ments. Nevertheless, the fact that they were matched the relationship between configural processing, processing
makes it appropriate to comment on the differential capacity, and the Thatcher illusion. Given the surprising lack
trends across eyes and mouths. We suggest that the catego- of evidence for configural processing in the Thatcher illusion,
risation task used in Experiment 1 was approached much as a it may be that other reported configural face effects should be
recognition task, in which attention tends to be allocated to similarly explored.
eyes (Riby, Riby, & Reay, 2009; Smith et al., 2004). The The present experiments were designed to address the issue
same/different task used in Experiment 2 may have been of configurality in terms of supercapacity processing, and they
approached more as a pattern-matching exercise, such that cannot directly speak to questions of processing architecture
the preference to attend to the eyes was removed. (serial vs. parallel vs. coactive) or the stopping rule (self-
The present finding goes beyond the fact that the orientation terminating vs. exhaustive) (see Wenger & Townsend,
of whole faces is critical to enabling the perception of 2001). However, the lack of evidence for supercapacity in
grotesqueness in the Thatcher illusion. The present the present data suggests that a coactive architecture with
experiments addressed whether there is a positive interaction positive dependencies between feature channels (eyes and
between Thatcherised features, as measured by processing mouth) can be ruled out as a possible architecture for the
capacity, and showed that no more work is necessarily processing of the Thatcher illusion. For instance, a positively
achieved in response to full-Thatcher faces than the sum of dependent coactive architecture would give rise to capacity
the work achieved in response to individual features. coefficient values of up to 5 (see Townsend & Wenger, 2004).
These data are inconsistent with upright faces leading Future research should extend these findings such that the
to a configural superiority effect for the detection of results can speak directly to the processing architecture and
grotesqueness in the Thatcher illusion, but they are stopping rule, as well as to processing capacity. Such research
consistent with research demonstrating that effects that should also evaluate performance in tasks that require exhaus-
appear configural in nature can be accounted for by nonconfi- tive processing (e.g., Wenger & Townsend, 2006). We also
gural models (Fifić & Townsend, 2010). suggest that other tasks used as markers of configurality
The lack of supercapacity in processing Thatcherised should be explored in terms of processing capacity.
faces is apparently in contrast to the findings of Ingvalson
and Wenger (2005), who found supercapacity processing in
Author note A subset of the data presented here were also reported
their test of the dual-mode hypothesis. However, there was in Katherine Cornes’s PhD Thesis (2008, University of Southampton),
an important difference between their work and the present titled Spatial and Temporal Factors in the Development of Second
experiments. Here, we measured capacity for two configural Order Relational Processing. Cornes’s PhD was funded by the
Economic and Social Research Council. Thank you to Helen J.
changes that led to grotesqueness; hence, both changes were
Richards for her advice on the survivor function analyses, and to
of the same type, and combining the two changes would not Michael J. Wenger, James T. Townsend, Mario Fifić, Joshua
affect the type of information available. Ingvalson and Solomon, and Stefan Schweinberger for their comments on earlier
Wenger, in contrast, were measuring capacity for different drafts of the manuscript.
types of changes (featural and configural), to test whether
the two channels of information interact. The difference
between a single change (featural or configural) and both References
types of changes is not only in the amount of information,
but in the types of information available. This change in
Allison, P. D. (2010). Survival analysis using SAS: A practical guide
information type did not take place for the stimuli used here. (2nd ed.). Cary: SAS Institute Inc.
The results of the present study should be considered Bartlett, J. C., & Searcy, J. (1993). Inversion and configuration of
alongside those of Cornes et al. (2011). Cornes et al. faces. Cognitive Psychology, 25, 281–316.
Blaha, L. (2004, July). From nonsense to gestalt: The influence of
explored the Thatcher illusion within the terms of general
configural learning of processing capacity. Paper presented at the
recognition theory (GRT), using response frequencies rather 37th Annual Meeting of the Society for Mathematical Psychology,
than latencies. They demonstrated that the effect of inversion Ann Arbor, Michigan.
Atten Percept Psychophys (2012) 74:1475–1487 1487

Blaha, L. M., & Townsend, J. T. (2006, May). Configural learning Mullin, P. A., Egeth, H. E., & Mordkoff, J. T. (1988). Redundant-
fundamentally changes the visual information processing system. target detection and processing capacity: The problem of
Paper presented at the annual meeting of the Vision Sciences positional preferences. Perception & Psychophysics, 43,
Society, Sarasota, Florida. 607–610.
Boutsen, L., Humphreys, G. W., Praamstra, P., & Warbrick, T. (2006). Perry, L. A. (2008). The architecture of same–different face judgments.
Comparing neural correlates of configural processing in faces and Unpublished Undergraduate Thesis, Indiana University, Bloo-
objects: An ERP study of the Thatcher illusion. NeuroImage, 32, mington, IN.
352–367. Perry, L. A., Blaha, L., & Townsend, J. T. (2008, May). Reassessing
Bradshaw, J. L., & Wallace, G. (1971). Models for the processing and the architecture of same–different face judgments. Paper pre-
identification of faces. Perception & Psychophysics, 9, 443–448. sented at the annual meeting of the Vision Sciences Society, Fort
Cornes, K., Donnelly, N., Godwin, H. J., & Wenger, M. J. (2011). Myers, FL.
Perceptual and decisional factors influencing the discrimination of Pomerantz, J. R., Sager, L. C., & Stoever, R. J. (1977). Perception of
inversion in the Thatcher illusion. Journal of Experimental wholes and of their component parts: Some configural superiority
Psychology. Human Perception and Performance, 37, 645–668. effects. Journal of Experimental Psychology. Human Perception
Cox, D. R. (1972). Regression models and life-tables. Journal of the and Performance, 3, 422–435.
Royal Statistical Society: Series B, 34, 187–220. Ratcliff, R. (1993). Methods for dealing with reaction time outliers.
Donnelly, N., & Hadwin, J. A. (2003). Children’s perception of the Psychological Bulletin, 114, 510–532. doi:10.1037/0033-
Thatcher illusion: Evidence for development in configural face 2909.114.3.510
processing. Visual Cognition, 10, 1001–1017. Riby, D. M., Riby, L. M., & Reay, J. L. (2009). Differential sensitivity
Donnelly, N., Zürcher, N. R., Cornes, K., Snyder, J., Naik, P., Hadwin, J., to rotations of facial features in the Thatcher illusion. Psycholog-
& Hadjikhani, N. (2011). Discriminating grotesque from typical ical Reports, 105, 721–726.
faces: Evidence from the Thatcher illusion. PLoS One, 6, e23340. Rotshtein, P., Malach, R., Hadar, U., Graif, M., & Hendler, T.
doi:10.1371/journal.pone.0023340 (2001). Feeling or features: Different sensitivity to emotion
Eriksen, C. W. (1988). A source of error in attempts to distinguish in high-order visual cortex and amygdala. Neuron, 32, 747–
coactivation from separate activation in the perception of redundant 757.
targets. Perception & Psychophysics, 44, 191–193. doi:10.3758/ Smith, M. L., Gosselin, F., & Schyns, P. G. (2004). Receptive fields for
BF03208712 flexible face categorizations. Psychological Science, 15, 753–761.
Fifić, M., & Townsend, J. T. (2010). Information-processing alternatives Spezio, M. L., Adolphs, R., Hurley, R. S. E., & Piven, J. (2007).
to holistic perception: Identifying the mechanisms of secondary- Analysis of face gaze in autism using “Bubbles”. Neuropsy-
level holism within a categorization paradigm. Journal of Experi- chologia, 45, 144–151. doi:10.1016/j.neuropsychologia.
mental Psychology: Learning, Memory, and Cognition, 36, 1290– 2006.04.027
1313. doi:10.1037/a0020123 Stürzel, F., & Spillmann, L. (2000). Thatcher illusion: Dependence on
Grice, G. R., Canham, L., & Boroughs, J. M. (1984). Combination rule angle of rotation. Perception, 29, 937–942.
for redundant information in reaction time tasks with divided Thompson, P. (1980). Margaret Thatcher: A new illusion. Perception,
attention. Perception & Psychophysics, 35, 451–463. 9, 483–484. doi:10.1068/p090483
Homa, D., Haver, B., & Schwartz, T. (1976). Perceptibility of schema- Townsend, J. T., & Ashby, F. G. (1978). Methods of modeling capacity
tic face stimuli: Evidence for a perceptual Gestalt. Memory & in simple processing systems. In N. J. Castellan & F. Restle
Cognition, 4, 176–185. (Eds.), Cognitive theory (Vol. 3, pp. 199–239). Hillsdale:
Ingvalson, E. M., & Wenger, M. J. (2005). A strong test of the dual- Erlbaum.
mode hypothesis. Perception & Psychophysics, 67, 14–35. Townsend, J. T., & Nozawa, G. (1995). Spatio-temporal properties of
Leder, H. (1996). Line drawings of faces reduce configural processing. elementary perception: An investigation of parallel, serial, and
Perception, 25, 355–366. coactive theories. Journal of Mathematical Psychology, 39, 321–
Leder, H., & Bruce, V. (2000). When inverted faces are recognized: 359. doi:10.1006/jmps.1995.1033
The role of configural information in face recognition. Quarterly Townsend, J. T., & Wenger, M. J. (2004). A theory of interactive
Journal of Experimental Psychology, 53A, 513–536. parallel processing: New capacity measures and predictions for
Lewis, M. B. (2001). The lady’s not for turning: Rotation of the a response time inequality series. Psychological Review, 111,
Thatcher illusion. Perception, 30, 769–774. 1003–1035. doi:10.1037/0033-295X.111.4.1003
Maurer, D., Grand, R. L., & Mondloch, C. J. (2002). The many faces of Wenger, M. J., & Gibson, B. S. (2004). Using hazard functions to
configural processing. Trends in Cognitive Sciences, 6, 255–260. assess changes in processing capacity in an attentional cuing
Menneer, T., Barrett, D. J. K., Phillips, L., Donnelly, N., & Cave, K. R. paradigm. Journal of Experimental Psychology. Human Perception
(2007). Costs in searching for two targets: Dividing search across and Performance, 30, 708–719.
target types could improve airport security screening. Applied Wenger, M. J., & Townsend, J. T. (2000). Basic response time tools for
Cognitive Psychology, 21, 915–932. doi:10.1002/acp.1305 studying general processing capacity in attention, perception, and
Miller, J. (1982). Divided attention: Evidence for coactivation with cognition. The Journal of General Psychology, 127, 67–99.
redundant signals. Cognitive Psychology, 14, 247–279. Wenger, M. J., & Townsend, J. T. (2001). Faces as Gestalt stimuli:
doi:10.1016/0010-0285(82)90010-X Process characteristics. In M. J. Wenger & J. T. Townsend (Eds.),
Mordkoff, J. T., & Egeth, H. E. (1993). Response time and accuracy Computational, geometric, and process perspectives on facial
revisited: Converging support for the interactive race model. cognition: Contexts and challenges (pp. 229–284). Mahwah:
Journal of Experimental Psychology. Human Perception and Erlbaum.
Performance, 19, 981–991. doi:10.1037/0096-1523.19.5.981 Wenger, M. J., & Townsend, J. T. (2006). On the costs and benefits of
Mordkoff, J. T., & Yantis, S. (1991). An interactive race model of faces and words: Process characteristics of feature search in
divided attention. Journal of Experimental Psychology. Human highly meaningful stimuli. Journal of Experimental Psychology.
Perception and Performance, 17, 520–538. doi:10.1037/0096- Human Perception and Performance, 32, 755–779. doi:10.1037/
1523.17.2.520 0096-1523.32.3.755

Potrebbero piacerti anche