PreprintPDF Available
Preprints and early-stage research may not have been peer reviewed yet.

Abstract

The brain calibrates itself based on the past stimulus diet, which makes frequently observed stimuli appear as typical (as opposed to uncommon stimuli, which appear as distinctive). Based on predictive processing theory, the brain should be more prepared for typical exemplars, because these contain information that has been encountered frequently, and allow to economically represent items of that category. Thus, one could ask whether predictability and typicality of visual stimuli interact, or rather act in an additive manner. We adapted the design by Egner and colleagues (2010), who used cues to induce expectations about stimulus category (face vs. chair) occurrence during an orthogonal inversion detection task. We measured BOLD responses with fMRI in 35 participants. First, distinctive stimuli always elicited stronger responses than typical ones in all ROIs, and our whole-brain directional contrasts for the effects of typicality and distinctiveness converge with previous findings. Second and importantly, we could not replicate the interaction between category and predictability reported by Egner et al. (2010), which casts doubt on whether cueing designs are ideal to elicit reliable predictability effects. Third, likely as a consequence of the lack of predictability effects, we found no interaction between predictability and typicality in any of the four tested regions (bilateral fusiform face areas, lateral occipital complexes) when considering both categories, nor in the whole brain. We discuss the issue of replicability in neuroscience and sketch an agenda for how future studies might address the same question.
Investigating the neural effects of typicality and predictability for face and object
stimuli
1
1Investigating the neural effects of typicality and predictability for face and object
2stimuli
3 Linda Ficco1,2,3*, Chenglin Li 2,4, Jürgen M. Kaufmann1, Stefan R. Schweinberger1,3 Gyula Z.
4 Kovács2,
51 Department of General Psychology and Cognitive Neuroscience, Friedrich Schiller
6 University, Jena, Germany
72 Department of Biological Psychology and Cognitive Neurosciences, Friedrich Schiller
8 University, Jena, Germany
93 International Max-Planck Research School for the Science of Human History, Jena,
10 Germany
11 4 School of Psychology, Zhejiang Normal University, Jinhua, China
12 *Corresponding author
13
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
2
14 Abstract
15 The brain calibrates itself based on the past stimulus diet, which makes frequently observed
16 stimuli appear as typical (as opposed to uncommon stimuli, which appear as distinctive). Based
17 on predictive processing theory, the brain should be more “prepared” for typical exemplars,
18 because these contain information that has been encountered frequently, and allow to
19 economically represent items of that category. Thus, one could ask whether predictability and
20 typicality of visual stimuli interact, or rather act in an additive manner. We adapted the design
21 by Egner and colleagues (2010), who used cues to induce expectations about stimulus category
22 (face vs. chair) occurrence during an orthogonal inversion detection task. We measured BOLD
23 responses with fMRI in 35 participants. First, distinctive stimuli always elicited stronger
24 responses than typical ones in all ROIs, and our whole-brain directional contrasts for the effects
25 of typicality and distinctiveness converge with previous findings. Second and importantly, we
26 could not replicate the interaction between category and predictability reported by Egner et al.
27 (2010), which casts doubt on whether cueing designs are ideal to elicit reliable predictability
28 effects. Third, likely as a consequence of the lack of predictability effects, we found no
29 interaction between predictability and typicality in any of the four tested regions (bilateral
30 fusiform face areas, lateral occipital complexes) when considering both categories, nor in the
31 whole brain. We discuss the issue of replicability in neuroscience and sketch an agenda for
32 how future studies might address the same question.
33
34
35 Keywords: Predictive coding; typicality; norm-based coding; fMRI; vision; perceptual space;
36 prototype-based models; exemplar-based models;
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
3
37 1. Introduction
38 1.1 Perceptual Spaces in Visual Perception
39 If one asks you to imagine the picture of a chair, chances are that it would be made of
40 wood, have four legs, be rather symmetrical, have a simple backrest and maybe a pillow.
41 Within any visual category, some exemplars are more representative than others, and are
42 perceived as “typical” (1). The perception of typicality depends on the fact that exemplars are
43 not represented independently in our mind (and brain). Any chair exemplar is remembered and
44 processed in relation to other similar-looking items, seen in similar contexts, with the same
45 function, size, material, and so on. Perceptual categories, such as “chairs”, can be thought of
46 as low-dimensional perceptual spaces of various exemplars (2–4), with minimal differences
47 and maximal similarities across exemplars ((1), p. 491).
48 In such models, typical exemplars share most features with each other (5), have more
49 average features (6), stand out of the category less (7), and summarize its dimensions best (5,8).
50 Importantly, the perception of typicality is flexible: it is influenced by the current perceptual
51 diet of the observer (9) and its stability depends on the duration and amount of previous
52 exposure (see (1) for a discussion).
53 Faces are thought to be represented in such space (6,10,11). We are exposed to
54 thousands of faces throughout our life (12), we develop expertise for this category (13,14). Our
55 face diet shapes the prototype in reference to which new faces are encoded constantly (15–
56 18), and people with allegedly comparable face spaces tend to form coherent typicality
57 judgements about the same faces (19).
58 A long-standing debate exists regarding different face-space models (11,20,21). In
59 exemplar-based models faces are encoded based on their similarity, at relative distances from
60 each other (6,22). Conversely, in norm-based models, faces are encoded with respect to a
61 weighted average of all seen faces (23). Notably, norm-based coding appears to extend beyond
62 facial stimuli to non-face objects as well (24–27). When comparing the two models explicitly,
63 the norm-based model seems to account for a number of findings e.g., responses in the
64 fusiform face area – better than the exemplar-based model (28). However, note that other works
65 comparing these two models, and using visual stimuli for which participants have acquired
66 expertise (e.g., faces, real objects and abstract shapes), seem to find support for both
67 (21,24,29,30).
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
4
68 1.2 Behavioural and Neural Effects of Typicality
69 Despite different existing operationalizations of typicality (e.g., different kinds of
70 ratings, via morphing, etc…), its effects are relatively uncontroversial: Neural and behavioural
71 data generally show that typical stimuli are easier to process (that is, performing mental
72 operations, such as detection or classification, on them) but harder to encode (i.e., store in a
73 mental representation). Behaviourally, typical stimuli are categorized faster and more
74 accurately than distinctive ones (24,25,31–34). Moreover, typical stimuli are detected faster
75 than distinctive ones (35), as suggested by detection advantages for own-“race” (typical), as
76 compared to other-“race” faces (6,36). However, typicality hampers the recognition of specific
77 exemplars: distinctive faces are recognized more accurately (37,38,47,39–46) and faster
78 (34,37–39,46) but see (48) than typical faces. This is possibly due to poorer encoding of typical
79 stimuli (49), which are more similar to each other and can be confused easier with each other,
80 leading to false positive answers by worse pattern separation processes (50).
81 Several neuroimaging techniques have been used to map perceptual spaces (reviewed
82 in (51). Neuroimaging findings suggest that typical stimuli require i) smaller brain metabolic
83 changes and ii) reduced configural processing, as compared to distinctive stimuli. Specifically,
84 fMRI studies report increased brain responses to distinctive stimuli, both in the visual
85 (28,52,53) and in the auditory modality (54,55). Conversely, electrophysiological studies tend
86 to use event-related potentials (ERPs). In ERP studies, face typicality consistently affects the
87 amplitude of the occipito-temporal P200 component, with larger amplitudes for typical, as
88 compared to distinctive stimuli (10,56–58). Object typicality potentially affects occipito-
89 temporal ERPs even at latencies < 200 ms (59); studies using objects or animal images report
90 larger P200 in frontal regions to distinctive as compared to typical objects (31) and shorter
91 latencies of the P300 component for typical stimuli (60). These findings parallel reaction time
92 data (60) and could indicate that typical stimuli require less attention and lower feature
93 encoding (61).
94 Importantly, and although it should be kept in mind that the precise definition of
95 typicality may vary across individual studies, neural and behavioural typicality effects appear
96 to be related, and their relationship cannot be explained by the mere physical similarity of
97 stimuli (52).
98
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
5
99 1.3 Typicality Effects under the Predictive Coding Framework: The Current Study
100 Visual perception does not only capitalize on dimensionality reduction (as discussed
101 above), but also on the prediction of upcoming sensory inputs (62). According to the theory of
102 predictive processing (63–65), the brain attempts to predict the input received from lower
103 processing levels, by generating an internal model of the input at each level of stimulus
104 processing (be it single neurons, neural populations, or network hubs) (66,67). The degree to
105 which this internal model differs from the incoming input represents the prediction error (i.e.,
106 the mismatch between actual and predicted input; (68)), which is then transferred to higher
107 levels (67). These recursive model-error comparison loops tend to minimize prediction error
108 (69), and ultimately allow the individual to bear the most accurate and up-to-date models of
109 the sensory world. In other words, perception is not a passive process, but an active attempt of
110 the brain to guess the latent causes of a sensory input best, informed by prior knowledge
111 (70,71).
112 Visual prototypes seem to be just this: representations “summarizing" stimuli from a
113 particular category best, based on those stimuli we frequently experienced in the past.
114 Potentially, this theory elegantly explains the neural and behavioural effects of stimulus
115 typicality as well: typical stimuli elicit less prediction error than distinctive ones when
116 compared to a prototype, thus leading to faster detection and categorization, and lower neural
117 responses in regions processing stimulus configuration and category. However, to date
118 systematic research on the combined effects of predictability and typicality is missing. To
119 address this important theoretical and empirical gap, one should test to which degree typicality
120 and predictability interact, or act in an additive manner. An interaction between predictability
121 and typicality would imply a common neural mechanism underlying both effects, whereas an
122 additive effect would be more consistent with separable mechanisms to compute stimulus
123 probability and typicality (a logic similar as that used in (72,73). Accordingly, if two processes
124 are independent, the expression of one effect should not change for different levels of the other
125 (i.e., the two effects should just add up). Instead, if there is an additional increase in one effect
126 for one level of the other that is not explained by a simple addition of the two effects, one can
127 infer an (at least partial) dependency of the mechanisms underlying the two effects.
128 To test this, we presented face or chair stimuli, preceded by a cue which signalled the
129 participants the probability of the occurrence of a stimulus category (“predictability” here thus
130 indicates expected temporal association between a stimulus category and a cue preceding the
131 stimulus). Importantly, orthogonal to this predictability modulation, we also manipulated
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
6
132 stimulus typicality by presenting typical and distinctive stimulus exemplars (so, typicality here
133 refers to the distance of each stimulus from an average exemplar of the same category see
134 Section 2.2). We replicated a design that produced an interaction between category and
135 predictability in a previous study (74), and that could allow us to elegantly test the presence of
136 an interaction. Another important goal of the present work is to assess the replicability of this
137 effect, especially since a recent, large study (75) showed it is challenging to produce neural
138 predictability effects with cueing, despite effective behavioural training.
139
140 2. Materials and Methods
141 2.1 Participants
142 We recruited thirty-five adult participants (22 females, one diverse; mean age = 24.4
143 years; SD = 4.0 years; two left-handed, one ambidextrous), with normal or corrected-to-normal
144 vision and with no reported neurological condition (one participant reported a diagnosis of
145 Asperger Syndrome, their data were retained in the analysis). The measurements were
146 performed between May and August 2022. The main experimenter had access to identifying
147 information from participants and took care of making their data anonymous for the analyses.
148 We ensured that each had > 10 years-long exposure to Caucasian/European/White faces. The
149 sample size was determined by a power analysis with the R package Superpower (76). We
150 calculated the number of participants necessary to detect a medium effect size (f = 0.25) for
151 the interaction of cue-induced predictability and category, to achieve a power of 0.80 at the
152 standard .05 alpha error probability. This was calculated based on the mean beta estimates in
153 different conditions reported in Figure 3 of (74). We chose this effect for our power analysis
154 because we had no expectations about the size of the interaction of interest, so we preferred to
155 calibrate our sample size to target a known effect. This led to a sample size that nearly doubled
156 the sample reported in (74), who recruited 17 participants. Participants could choose between
157 monetary compensation and a small 3D model of their own brain for their participation. Before
158 the experiment, all participants received information about the experimental procedures and
159 provided their written informed consent. The ethics committee of the Friedrich-Schiller-
160 Universität Jena approved the experimental protocol (Reg. No. FSV 22/086), and the study was
161 conducted in accordance with the guidelines of the Declaration of Helsinki. The pre-
162 registration for this study can be found
163 at:https://osf.io/axstg/?view_only=13f0e8794885499d805e6649670b2f13.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
7
164
165 2.2 Stimuli
166 In addition to faces we chose images of chairs as control stimuli, as they show
167 approximate vertical symmetry and a clear upright direction. Both these features seem to be
168 important in norm-based coding of objects (77). As cues we used geometrical shapes of
169 different colours (a green, a blue, and a yellow frame, with similar area, brightness, and
170 saturation levels). All stimuli were presented by Psychtoolbox-3 (MATLAB-based; (78,79)).
171 We manipulated perceived typicality of faces and chairs as follows. For faces, we
172 followed previous works as regards the stimulus database and the manipulation approach
173 (10,80–82). In detail, we created three-dimensional face stimuli using DI3Dcapture™
174 (Dimensional Imaging, Glasgow, UK). Each face had been photographed by four cameras
175 simultaneously, and the images were interpolated to create a three-dimensional (3D) object.
176 Using DI3Dview™, we generated caricatures and anti-caricatures (both in shape and texture),
177 by extrapolating each individual veridical 3D file with the morphing tool with respect to a
178 gender-matched average. Finally, because the 3D camera system permits to extract images
179 from various viewing angles, we produced our stimuli by systematically tilting the individual
180 faces according to a set of ten camera angles (see (80) for a more detailed description of this
181 procedure). Based on a previous rating study, we selected 24 face identities (half female and
182 half male). We subjected each face to both caricaturing (shifting both face shape and texture +
183 0.33 units on the face trajectory, opposite to the face norm) and anti-caricaturing (shifting both
184 face shape and texture + 0.33 units on the face trajectory, in the direction of the face norm).
185 For each participant, we selected only one caricaturing version per identity (counterbalanced
186 across participants), leading to 12 caricatured (distinctive) and 12 anti-caricatured (typical)
187 identity for each participant. Throughout the manuscript we use the term “typical” to refer to
188 both faces morphed towards the average and to chairs rated as typical. Conversely, the term
189 “distinctive” refers both to faces morphed in the opposite direction with respect to the average
190 and to chairs rated as distinctive. Instead, “anti-caricatured” and “caricatured” refer to the
191 procedure applied to make faces typical and distinctive, respectively. Since we wanted to avoid
192 image-dependent identity perception (i.e., the same picture of Identity 4 is shown throughout
193 the experiment), we produced 10 different images for each identity. We used Adobe
194 Photoshop™ to tilt each face on 10 pre-set angles, resulting in 240 stimuli in total for
195 participant for each task run.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
8
196 Colour images of chairs were downloaded from the Internet and were subjected to
197 standard pre-processing to ensure similar image size, quality, and a homogeneous background.
198 No chair contains written text or recognizable logos. To make the chairs comparable to face
199 images, we used 10 images per chair, each showing the image on a comparable set of angles.
200 We manipulated typicality by selecting the 12 most typical and the 12 most distinctive chair
201 images, as rated by a separate group of 10 participants (Cronbach’s alpha = 0.991).
202 Both face and chair stimuli were presented in black and white, to avoid confounding
203 effects due to the wider range of colours of chairs, as compared to faces. See Figure 1 for
204 examples of face and chair stimuli.
205 INSERT FIGURE 1 HERE
206 Figure 1. Stimulus examples. We show here the same face identity in its caricatured and anti-
207 caricatured version for comparison. Note that each identity was shown to each participant in
208 either of the two versions only. Left: a typical (anti-caricatured) face and a typical chair. Right:
209 a distinctive (caricatured) face and a distinctive chair.
210
211 2.3 Experimental Design and Procedure
212 To manipulate predictability of stimulus category (face or chair), a green frame
213 preceded a face in 75% of the trials, and a chair in the remaining 25% of trials. Conversely, a
214 blue frame preceded a chair in 75% of the trials and a face in the remaining 25%. As suggested
215 by (83) and following (74), we included a neutral, uninformative condition, whereby a yellow
216 frame preceded faces or chairs equally often (50% of the trials). Throughout the task,
217 participants saw 24 identities (12 faces + 12 chairs), in 10 camera angles. Participants
218 performed four task runs in total (two participants performed only three runs due to technical
219 issues). The order of trials within each experimental run was randomized. Each trial started
220 with a fixation cross, which also separated trials (2000-4000 ms, plus a blank if participants
221 responded, to ensure that the duration of one trial is an integer multiple of our TR = 2),
222 followed by a cue (250 ms), and a target stimulus (750 ms). To ensure participants’ attentional
223 focus, in 10% of these trials the stimulus was inverted (but otherwise followed the same cue
224 contingencies as used for upright stimuli), and participants performed an orthogonal detection
225 task of these trials by pressing a key to inverted stimuli with their right index finger on a MRT-
226 compatible keyboard. The experiment included a total of 528 trials, divided in 4 runs of 132
227 trials each (39 usable, non-target trials per condition in each run). Target trials were excluded
228 from analyses. Participants were informed about the cue contingencies, but we specified that
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
9
229 these were not task relevant. We did so with the aim to replicate the design by (74), who
230 instructed their participants in a similar way. Before completing the task, participants were
231 familiarized with the procedure and the contingencies through a short practice session (~2.5
232 minutes) outside the scanner.
233 After completing the fMRI measurement, participants completed a short questionnaire
234 in which they reported the degree to which they experienced any difficulties during scanning,
235 paid attention to the contingencies, and thought the faces and chairs looked typical or
236 distinctive (cf. Supplementary Materials).
237 INSERT FIGURE 2 HERE
238 Figure 2: Graphical representation of the experimental conditions.
239
240 2.4 Imaging Parameters and Data Analysis
241 We used a 3-Tesla MR scanner (Siemens MAGNETOM Prisma fit, Erlangen,
242 Germany) with a 20-channel head coil to record BOLD responses to our manipulations. T1-
243 weighted high-resolution 3D anatomical images were acquired with an MP-RAGE sequence
244 (192 slices; TR = 2300 ms; TE = 3.03 s; flip angle = 90°; 1 mm isotropic voxel size). As for
245 the four functional runs, T2*-weighted fMRI-images were collected with a multi-band EPI
246 sequence (MB acceleration factor = 8) under the following parameters: 34 slices; FOV = 204
247 x 204 mm2; TR = 2000 ms; TE = 30 ms; flip angle = 90°; 2 mm isotropic voxel size. Similarly
248 to a previous study (84), we used SPM12 (Welcome Department of Imaging Neuroscience,
249 London, UK), based on MATLAB version R2020a, to pre-process data.
250 The experiment was based on an event-related design. Functional images were slice-
251 timed, realigned (the structural image was realigned to a mean image computed from the
252 functional series, and co-registered to structural scans). We then normalized the images to the
253 MNI-152 space, resampled to 2 x 2 x 2 mm resolution, and spatially smoothed using an 8-mm
254 Gaussian kernel. The convolution of a reference hemodynamic response function (HRF) with
255 box cars, representing the onsets and durations of the experimental conditions, was used to
256 define the regressors for a general linear model analysis of the data. For each of the
257 experimental conditions we modelled the HRF, synchronised to the onset of the trial. Low
258 frequency components were excluded from the model using a high-pass filter with 128 s cut-
259 off. Variance which could be explained by the previous scans was excluded using an
260 autoregressive AR(1) model, and movement related variance was accounted for by the spatial
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
10
261 parameters resulting from the realignment procedure. The resulting regressors were fitted to
262 the observed functional time series. For the random effects analysis, the contrast estimates
263 entered a simple t-test or an F-test at the second level (see Table S3 for the regressor weights
264 in the whole-brain analysis). Results were thresholded at p < 0.05 (voxelwise, uncorrected),
265 with a cluster correction for multiple comparisons at FWE < 0.05. For visualization, the
266 thresholded t-images were superimposed onto a standard template available on MRIcroGL
267 (https://www.nitrc.org/projects/mricrogl/). We extracted the peaks of maximal activation and
268 their respective anatomical labels using the Anatomy Toolbox implemented in (93) and the
269 Talairach Daemon Client (94,95).
270 Additionally, participants performed a 6-minutes functional localizer task, so that we
271 could isolate regions of interest (ROIs). Faces, everyday objects, and Fourier-randomized noise
272 images were presented (4 Hz; 230ms exposition time; 20 ms ISI) in blocks of 10 s, interrupted
273 by breaks of 10 s and repeated five times. Each block included 40 images (size: 400 x 400
274 pixels with a grey background). Participants were instructed to observe the images and
275 maintain their attentional focus on the screen. The fusiform face area (FFA; (85,86)) and the
276 lateral occipital complex (LOC; (87,88)) were isolated. The object-selective area LO (87,89)
277 corresponded to the posterior dorsal portion of the lateral occipital complex (LOC, (90)). We
278 determined the location of the FFA in individual participants by contrasting face and object
279 blocks (face and noise blocks, if the former led to no significant voxels) and established as the
280 local maximum from the t-maps with a threshold of
𝑝
𝐹𝑊𝐸
< 0.05 on the single-subject level. A
281 similar approach was taken to locate the LOC (objects > noise blocks, t-maps with a threshold
282 of
𝑝
𝐹𝑊𝐸
< 0.05), on the single-subject level. We report the individual MNI coordinates in Table
283 S1, and report the average (± SD) here: right FFA (
𝑥
𝑀
= 40.6,
𝑥
𝑆𝐷
= ± 4.7;
𝑦
𝑀
= -52.9,
𝑦
𝑆𝐷
=
284 ± 7.7;
𝑧
𝑀
= -18.6,
𝑧
𝑆𝐷
= ± 3.9), left FFA (
𝑥
𝑀
= -38.9,
𝑥
𝑆𝐷
= ± 3.5;
𝑦
𝑀
= -52.4,
𝑦
𝑆𝐷
= ± 6.9;
𝑧
𝑀
=
285 -18.9,
𝑧
𝑆𝐷
= ± 4.0), right LOC (
𝑥
𝑀
= 40.1,
𝑥
𝑆𝐷
= ± 4.6;
𝑦
𝑀
= -76.1,
𝑦
𝑆𝐷
= ± 8.2;
𝑧
𝑀
= -3.6,
𝑧
𝑆𝐷
286 = ± 6.4), left LOC (
𝑥
𝑀
= -43.1,
𝑥
𝑆𝐷
= ± 5.1;
𝑦
𝑀
= -78.3,
𝑦
𝑆𝐷
= ± 6.6;
𝑧
𝑀
= -3.5,
𝑧
𝑆𝐷
= ± 5.7).
287 Areas matching anatomical criteria were considered as their appropriate equivalents on the
288 single subject level. A time-series of the mean voxel value within a 4 mm radius sphere around
289 the local peak was extracted from our event related sessions using finite impulse response (FIR)
290 models (91) for each area and participant separately. As for the main task analysis, we
291 convolved our data with a reference HRF using boxcars, to define the regressors for a general
292 linear model using MarsBaR 0.44 toolbox for SPM 12 (92). The peak BOLD values
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
11
293 (corresponding to the third TR post-stimulus onset) were extracted from the four event-related
294 runs while trials with upside-down stimuli were excluded from the analysis. We analysed only
295 upright trials using repeated-measures ANOVAs in each ROI with cue-induced Predictability
296 (3, high-face expectation, uninformative, high-chair expectation), Category (2, faces and
297 chairs) and Typicality (2, typical and distinctive) as within-subject factors. We were especially
298 interested in the main effects of typicality (both overall and separate by category), the
299 interaction between category and predictability (replication of (74)’s results) and, critically, the
300 interaction between typicality and predictability. These analyses were performed on JASP
301 (Version 0.16.3). All multiple comparisons of post-hoc tests were corrected with Holm’s
302 method.
303
304
305 3. Results
306 3.1 Behavioural Results
307 Participant's accuracy was close to ceiling (M = 0.99 proportion correct responses, SD
308 = 0.01, response time: M = 543 ms, SD = 77 ms). This suggests that they attended to the task
309 well, and that the task was easy to perform. Due to the small number of inverted trials, we did
310 not perform statistical analysis, but note that numerically, participants were faster for the
311 detection of inverted faces than for inverted chairs, M = 525.4 vs. 553.4 ms, respectively. All
312 participants confirmed that experimental faces were similar to those encountered throughout
313 their life. On average participants reported only relatively low attention to cue contingencies
314 (M = 0.77 on a scale from 0 to 3; see Table S2 for details) and we detected no effects of
315 predictability on RT (p = 0.1692). Note that participants were instructed that cue contingencies
316 were irrelevant to task completion. Finally, the presence of a typicality manipulation was
317 subjectively noticed for chairs more often than for faces, M = 20/35 vs. 5/35 participants,
318 respectively. We exploratively investigated behavioural effects of typicality, but found no
319 differences between typical and distinctive stimuli, neither on RTs (p = .163) nor in accuracy
320 (F(1, 416) = 1.237, p = .267).
321
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
12
322 3.2 Neuroimaging Results
323 3.2.1 Regions-of-Interest
324 We found prominent effects of Category and Typicality in all our ROIs (Figure 3). Here
325 we report the average between left and right hemisphere, see the Supplementary Materials the
326 results for the individual ROIs. Considering all participants, on a total of 12600 non-target trials
327 (counted considering all participants together), only 32 were false alarms. Since this number is
328 low, and we did not have reasons to expect these to affect any of our experimental conditions
329 differently, we did not remove them from brain data analysis. In the FFA the main effect of
330 Category was significant (F(1,28) = 29.058, p < .001,
𝜂
𝑝
= 0.509), with faces eliciting larger
331 responses than chairs (p = .001). The effect of Category was also significant in the LOC
332 (F(1,33) = 103.146, p < .001,
𝜂
𝑝
= 0.763), where chairs elicited larger responses than faces.
333 Considering that these regions were selected based on the functional localizer and extensive
334 previous literature, these results are expected, and can be seen as an indicator of data quality.
335 More interesting are the effects of Typicality over the four ROIs (all ps < .001), which were at
336 times comparably sizable as those of Category (e.g., right FFA,
𝜂
𝑝
= 0.612; range
𝜂
𝑝
s = 0.257
337 - 0.612). The effect of Typicality was significant and in the same direction even when
338 considering only the stimulus Category for which each ROI is specialized (i.e., faces for FFAs
339 and chairs for LOCs; all
𝑝𝑠
𝐻𝑜𝑙𝑚
< .005; see Figure S1). Remarkably, the hypothesized
340 interactions between stimulus Typicality and Predictability failed to reach significance (cf.
341 Table 2 for details). We additionally analysed this interaction when only considering the
342 preferred stimulus category of the ROIs (faces for FFA, chairs for LOC). Again, the
343 interactions did not reach significance, although in the LOC we note a trend (p = .081), with
344 distinctive chairs eliciting larger responses than typical chairs especially in the uninformative
345 condition. In line with (74), we found no main effect of Predictability in any ROI. We also
346 could not replicate the interaction between Category and Predictability reported by (74) either,
347 and no other interactions reached significance. We additionally report the Bayes Factor for the
348 main models we tested in the FFAs and the LOCs. In both cases, the model with the highest
349 evidence in favour of the alternative hypothesis was that including the two main effects of
350 Category and Typicality but not their interaction (
𝐵𝐹
10
= 1.000 in the FFAs, and
𝐵𝐹
10
= 1.000
351 in the LOCs). These values can be interpreted as weak evidence in favour of the described
352 model (96). Additionally, in both cases the model with the strongest evidence in favour of the
353 null hypothesis is that including solely the main effect of Predictability (
𝐵𝐹
10
=
5.363
in the
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
13
354 FFAs, and
𝐵𝐹
10
=
7.147
13
in the LOCs). The analysis for our model of interest, including the
355 main effects of Category, Typicality and Predictability, plus the interactions
356 Predictability*Typicality, returned strong support in favour of the null hypothesis both in the
357 FFAs (
𝐵𝐹
10
= 0.005) and in the LOCs (
𝐵𝐹
10
= 0.028) (96). Finally, the model including
358 Category, Predictability and their interaction in the FFAs, which would correspond to the effect
359 reported in (74) received strong evidence in favour of the null hypothesis (
𝐵𝐹
10
=
2.234
).
360 Overall, these analyses are consistent with the effects calculated under a frequentist framework
361 - although the evidence in favour of our two main effects remains weak and point to the absence
362 of our interaction of interest.
363
364 INSERT FIGURE 3 HERE
365
366 Figure 3: Results of the ROI analysis. Brain responses to typical and distinctive stimuli of
367 the two categories to facilitate the comparison with Figure 3 in (74). High_Face_Pred: cue that
368 highly predicts a face; High_Chair_Pred: cue that highly predicts a chair; Uninformative: cue
369 that predicts either category equally; panel A: brain responses to typical and distinctive stimuli
370 in both left and right FFAs averaged; panel B: brain responses to typical and distinctive stimuli
371 in both left and right LOCs averaged. For a figure depicting brain responses in individual ROIs,
372 see Figure S1. ROIs shown on the right for illustration purposes only, the average locations of
373 the individually defined ROIs are indicated in Section 2.4. Error bars represent 95% confidence
374 intervals.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object stimuli
14
375
376 Table 2. Within-subjects ANOVAs results and contrast weights
377 Within-subjects ANOVAs with Predictability (3 levels), Category (2 levels) and Typicality (2 levels) as within-subjects factors. Results are presented both in each ROI singularly and in face- and
378 object-responsive regions averaged across hemispheres. Significant effects are boldened. Sum of squares of type III are reported. a Mauchly's test of sphericity indicates that the assumption of
379 sphericity is violated (p < .05).
Left FFA (N = 21)
Effect
Sum of Squares
(residuals)
Df (residuals)
Mean Square (residuals)
F
p
𝜂
2
𝑝
Predictability
0.005a (0.439)
2a (40)
0.003a (0.011)
0.24a
.792a
.012a
Category
0.977 (1.180)
1 (20)
0.977 (0.059)
16.56
< .001
.453
Typicality
0.077 (0.156)
1 (20)
0.077 (0.008)
9.82
.005
.329
Predictability * Category
0.029a (0.354)
2a (40)
0.014a (0.009)
1.62a
.211a
.075a
Predictability * Typicality
0.004 (0.294)
2 (40)
0.002 (0.007)
0.25
.780
.012
Category * Typicality
0.217 (0.363)
1 (20)
0.217(0.018)
0.01
.929
.004
Predictability * Category *
Typicality
0.030a (0.647)
2a (40)
0.015a (0.016)
1.93a
.403a
.044a
Right FFA (N = 27)
Effect
Sum of Squares
(residuals)
Df (residuals)
Mean Square (residuals)
F
p
𝜂
2
𝑝
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object stimuli
15
Predictability
0.016 (0.493)
2 (52)
0.008 (0.009)
0.87
.427
.032
Category
0.880 (0.988)
1 (26)
0.880 (0.038)
23.16
< .001
.471
Typicality
0.180 (0.114)
1 (26)
0.180 (0.004)
40.97
< .001
.612
Predictability * Category
0.009 (0.468)
2 (52)
0.005 (0.009)
0.52
.599
.020
Predictability * Typicality
0.003 (0.296)
2 (52)
0.001 (0.006)
0.23
.797
.009
Category * Typicality
0.006 (0.128)
1 (26)
0.006 (0.005)
1.27
.271
.046
Predictability * Category *
Typicality
0.007 (0.382)
2 (52)
0.004 (0.007)
0.49
.614
.019
Left LOC (N = 29)
Effect
Sum of Squares
(residuals)
Df (residuals)
Mean Square (residuals)
F
p
𝜂
2
𝑝
Predictability
0.011 (0.482)
2 (56)
0.006 (0.009)
0.65
.529
.023
Category
1.379 (0.525)
1 (28)
1.379 (0.019)
73.52
< .001
.724
Typicality
0.090 (0.261)
1 (28)
0.090 (0.009)
9.70
.004
.257
Predictability * Category
0.010 (0.357)
2 (56)
0.005 (0.006)
0.80
.454
.028
Predictability * Typicality
0.002 (0.277)
2 (56)
0.001 (0.005)
0.24
.791
.008
Category * Typicality
0.017 (0.198)
1 (28)
0.017 (0.007)
2.47
.127
.081
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object stimuli
16
Predictability * Category *
Typicality
0.013 (0.549)
2 (56)
0.006 (0.010)
0.65
.526
.023
Right LOC (N = 31)
Effect
Sum of Squares
(residuals)
Df (residuals)
Mean Square (residuals)
F
p
𝜂
2
𝑝
Predictability
0.012 (0.588)
2 (60)
0.006 (0.010)
0.59
.557
.019
Category
2.340 (1.101)
1 (30)
2.340 (0.037)
63.79
< .001
.680
Typicality
0.218 (0.222)
1 (30)
0.218 (0.007)
29.49
< .001
.496
Predictability * Category
0.006 (0.411)
2 (60)
0.003 (0.007)
0.41
.666
.013
Predictability * Typicality
0.028 (0.387)
2 (60)
0.014 (0.006)
2.14
.126
.067
Category * Typicality
0.011 (0.123)
1 (30)
0.011 (0.004)
2.65
.114
.081
Predictability * Category *
Typicality
0.019 (0.509)
2 (60)
0.010 (0.008)
1.12
.333
.036
Average of left and right FFAs (N = 29)
Effect
Sum of Squares
(residuals)
Df (residuals)
Mean Square (residuals)
F
p
𝜂
2
𝑝
Predictability
0.002 (0.453)
2 (56)
0.000 (0.008)
0.12
.890
.004
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object stimuli
17
Category
1.170 (1.128)
1 (28)
1.170 (0.040)
29.06
< .001
.509
Typicality
0.125 (0.153)
1 (28)
0.125 (0.005)
22.87
< .001
.450
Predictability * Category
0.015 (0.334)
2 (56)
0.008 (0.008)
1.30
.281
.044
Predictability * Typicality
0.001 (0.285)
2 (56)
0.000 (0.005)
0.14
.867
.005
Category * Typicality
0.000 (0.204)
1 (28)
0.000 (0.007)
0.132
.719
.005
Predictability * Category *
Typicality
0.031 (0.428)
2 (56)
0.016 (0.008)
2.04
.193
.068
Average of left and right LOCs (N = 33)
Effect
Sum of Squares
(residuals)
Df (residuals)
Mean Square (residuals)
F
p
𝜂
2
𝑝
Predictability
0.010 (0.496)
2 (64)
0.005 (0.008)
0.64
.529
.020
Category
2.002 (0.621)
1 (32)
2.002 (0.019)
103.15
< .001
.763
Typicality
0.160 (0.211)
1 (32)
0.160 (0.007)
24.25
< .001
.431
Predictability * Category
0.002 (0.353)
2 (64)
0.001 (0.006)
0.19
.825
.006
Predictability * Typicality
0.010 (0.322)
2 (64)
0.005 (0.005)
0.96
.387
.029
Category * Typicality
0.007 (0.187)
1 (32)
0.007 (0.006)
1.25
.273
.037
Predictability * Category *
Typicality
0.026 (0.491)
2 (64)
0.013 (0.008)
1.69
.193
.050
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
18
381
382 3.2.2 Whole-brain analyses
Contrast
Local maxima
MNI standard space
x, y, z (t- or F-values)
Anatomical locations
(Brodmann area)
Cluster size
(voxels)
Chairs > Faces (t-contrast) p > .05, FWE
Cluster 1
26, -52, -10 (13.95)
Right fusiform gyrus (BA 19)
2763
42, -80, 4 (13.51)
Right middle occipital gyrus (BA 19)
32, -84, 14 (12.87)
Right middle occipital gyrus (BA 19)
22, -76, -10
Right lingual gyrus (BA 18)
18, -78, 44
Right precuneus (BA 7)
20, -76, 42
Right precuneus (BA 7)
48, -58, -6
Right middle occipital gyrus (BA 19)
Cluster 2
-36, -88, 8 (13.81)
Left middle occipital gyrus (BA 19)
2542
-30, -44, -12 (12.41)
Left fusiform gyrus (BA 37)
-32, -46, -10
Left fusiform gyrus (BA 37)
26, -58, -8
Right fusiform gyrus (BA 19)
-42, -78, -2
Left inferior occipital gyrus (BA 19)
-30, -74, 0
Left lingual gyrus (BA 19)
-44, -62, -4
Left middle temporal gyrus (BA 37)
-20, -76, -10
Left lingual gyrus (BA 18)
-22, -78, -8
Left lingual gyrus (BA 18)
Cluster 3
18, -62, 52 (6.31)
Right superior parietal lobule (BA 7)
32
22, -56, 56 (5.61)
Right precuneus (BA 7)
Faces > Chairs (t-contrast) p > .05, FWE
Cluster 1
54, -50, 8 (6.51)
Right superior temporal gyrus (BA 22)
64
Distinctive > Typical (t-contrast) p > .05, FWE
Cluster 1
34, -80, -10 (7.84)
Right fusiform gyrus (BA 19)
293
44, -80, 0 (7.64)
Right middle occipital gyrus (BA 19)
36, -78, -12
Right fusiform gyrus (BA 19)
42, -78, -2
Right inferior occipital gyrus (BA 19)
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
19
383 Table 1: Local maxima of activation for the contrasts of interest at the whole brain level. Note that t- or F-values
384 are provided for the peaks output by SPM12, whereas the other peaks have been located with the Anatomy
385 Toolbox. For them all, the third column reports the anatomical location assigned via the Anatomy Toolbox. The
386 raw output of the Anatomy Toolbox is provided as Supplementary Materials.
387
388 We investigated the main effects of Predictability (both by means a non-directional F
389 contrast and a directional t-contrast comparing high predictability conditions to the
390 uninformative one), Typicality and Category. We also investigated the interaction of interest
391 (Predictability * Typicality). The contrast faces > chairs revealed one cluster in the right
392 superior temporal gyrus (rSTG), whereas the opposite contrast returned activation in two large
393 bilateral clusters mainly encompassing the LOCs, the fusiform gyri, V1, V3, V5, extending to
394 the hippocampal cornu ammonis, and to the inferior and superior parietal lobules. The contrast
395 typical > distinctive generated a small cluster in the left superior parietal lobe, while the
396 opposite (distinctive > typical) revealed two symmetrical clusters which include the LOCs, the
397 fusiform gyri and the ventral portions of V2. We report the same effect estimated in faces and
398 chairs separately in Figure S2. The main effect of Predictability resulted in no suprathreshold
399 clusters (p < .05, FWE corrected), and relaxing the threshold to the conventionally accepted
400 level of p < .001 (uncorrected) resulted in two small clusters in the right hemisphere, one in the
401 anterior cingulate gyrus, and the other in the anterior insula. The directional contrast reflecting
Cluster 2
-34, -86, -6 (7.15)
Left inferior occipital gyrus (BA 18)
159
-24, -86, -12 (6.75)
Left inferior occipital gyrus (BA 19)
-28, -84, -12
Left fusiform gyrus (BA 19)
40, -72, -8 (6.21)
Right inferior occipital gyrus (BA 19)
-40, -78, -6
Left inferior occipital gyrus (BA 19)
-44, -80, -8
Left inferior occipital gyrus (BA 18)
Typical > Distinctive (t-contrast) p > .05, FWE
Cluster 1
-4, -64, 24 (6.08)
Left superior parietal lobule (BA 7a)
8
Predictability (F-contrast) p < .0001, uncorrected
Cluster 1
10, 4, 26 (17.79)
Right anterior cingulate gyrus (BA 33)
14
Cluster 2
-10, -2, 30 (17.92)
Left anterior cingulate gyrus (BA 33)
12
Highly informative cues > uninformative cue (t-contrast) p < .0001, uncorrected
Only two clusters < 3 voxels
Interaction between Predictability and Typicality (F-contrast) p < .0001, uncorrected
No significant results
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
20
402 high predictability conditions compared to the uninformative one revealed no significant results
403 (p < .05, FWE corrected). Setting a more lenient threshold at p < .0001, uncorrected resulted
404 only in a three-voxel cluster in the right anterior cingulate gyrus which we report, but refrain
405 from interpreting. The contrast relative to the interaction between Predictability and Typicality
406 was not significant.
407 INSERT FIGURE 4 HERE
408 Figure 4: Whole brain results (MNI standard space). From left to right: directional t-contrasts
409 of brain activation in response to faces > chairs (red) and chairs > faces /green); directional t-
410 contrasts of brain activations in response to distinctive > typical stimuli (dark blue) and typical
411 > distinctive (light green); non-directional F contrast assessing whether any of the predictability
412 levels differs from the others (yellow) and directional t-contrast comparing the two conditions
413 of high predictability to the uninformative condition (dark red). All contrasts are shown at an
414 alpha level of .05, with a family-wise correction for multiple comparisons.
415
416 4. Discussion
417 This study investigated combined effects of stimulus typicality and cue-induced
418 predictability. Our results confirm reports of increased responses to distinctive, as compared to
419 typical stimuli for both face and non-face stimuli. In contrast to our hypotheses, we neither
420 found robust effects of predictability, nor an interaction between predictability and typicality.
421 4.1 The Brain is Sensitive to Visual Typicality
422 We detected stronger brain responses to distinctive, as compared to typical stimuli in
423 all the tested ROIs, which confirms previous findings regarding face typicality encoding in the
424 FFA (28,52,53) and extends them to non-face stimuli and their processing in the LOC. Most
425 participants also seem to have noticed variations of typicality either in faces or chairs, when
426 explicitly asked after the experiment. Critically, the lack of interaction with stimulus category
427 indicated that these regions are sensitive to the distance of the stimulus from the prototype
428 regardless of category, which also appears from our separate analyses of typicality effects for
429 faces and chairs (Figure S2). Overall, these results align both with previous findings about the
430 neural effects of typicality (see Section 1.2) and may represent a neural signal encoding
431 distance from the prototype (also see (10), where increasing distance leads to decreasing P200
432 amplitude). Note that although this idea is more plausible when considering norm-based
433 models (6,18,26) it also can be reconciled with exemplar-based models of visual perception
434 (22,97,98). In this case, enhanced brain responses to distinctive stimuli might reflect distances
435 of these from the majority of the others (with no need of abstraction). These responses might
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
21
436 then represent a “rarity” signal, based on the fact that distinctive stimuli are also less frequent.
437 Finally, reduced responses to typical stimuli might also represent a phenomenon of neural
438 adaptation to statistical regularities associated to that particular category - when a face is
439 perceived, it is more likely to present certain features - in line with predictive accounts of vision
440 (99). However, as discussed below, the extent to which such adaptation concerns category-
441 space distances or rather low-level features remains an open question.
442 We also found increased brain responses to distinctive, as compared to typical stimuli
443 in two large bilateral clusters including the fusiform gyri, the middle and the inferior occipital
444 gyri in our whole-brain analysis. This aligns with the prediction that typicality processing
445 should occur in regions representing and tying visual features into coherent percepts, such as
446 the LOC (52). It is interesting to compare our univariate whole-brain results with those of a
447 multivariate study reporting that typical stimuli of two different categories are maximally
448 differentiated in the LOC, in which between-category boundaries are thus maximized (29). One
449 interpretation may be that the LOC maximally discriminates between categories by storing
450 prototypical representations, such that its responses to prototype-deviant stimuli can be seen as
451 a potential “surprise” signal. We additionally found a small significant cluster of voxels where
452 activity was larger for typical when compared to distinctive stimuli, corresponding to the
453 inferior parietal lobule (IPL), which is reminiscent of a cluster identified by (29) in a searchlight
454 analysis. In their study, neural patterns in response to distinctive (atypical) exemplars were
455 more differentiated in the caudal IPL. (29) speculate that this area facilitates recognition of
456 atypical items, achieved by relating these items to the respective category ((29), pp. 175-176).
457 Therefore, stronger activation to typical stimuli within the same area might reflect the
458 activation of prototypical representations necessary to establish the category to which the item
459 belongs. We also note that, while typicality information consolidated through the years (like
460 that to faces and common objects) seems to take place in the visual stream and the IPL, other
461 regions in the frontal lobe and the hippocampus might be involved in learning new prototypes,
462 from entirely new visual categories (24). The conceptual separation between structural and
463 functional typicality (1) is thus mirrored at the neural level.
464 As a cautionary note, the activation differences between typical and distinctive stimuli
465 may reflect differences in low-level features, rather than differences in the location within a
466 hypothetical category space. While it is difficult to determine the extent to which this is the
467 case in the current study and especially in the case of chair images, we note that we equalized
468 luminance and spatial frequency distribution in the case of faces, for which we still find
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
22
469 typicality effects in our ROI analysis. Moreover, (52) found no correlation between a neural
470 measure of physical typicality (height and angle of certain stimuli parts within the same
471 category) and another, based on subjective typicality (ratings). Although we acknowledge that
472 the stimuli used in (52) were more controlled in terms of low-level features and we note that
473 our typicality effects were larger for the less controlled chair stimuli than for faces, these
474 considerations indicate that typicality effects in our study may indeed be driven by category-
475 space structure and not only low-level features.
476
477 4.2 Failure to Produce Cue-induced Predictability Effects
478 An important finding is that we detected no main effect of cue-induced predictability
479 in the brain. This likely occurred due to the irrelevance of cue-stimulus contingencies during
480 the task. In fact, while a previous study with a similar cueing design reported lower brain
481 responses in ventral visual areas to strongly expected stimuli (100), (74) also did not find the
482 main effect of cue-based predictability in the fusiform face and parahippocampal place areas.
483 An explanation of the controversial results might be that, in our case and in (74), participants
484 were told explicitly that the cues were task irrelevant, and this corresponds with subjective
485 reports of low attention to cue contingencies in the present study. By contrast, in the study
486 above, participants were explicitly encouraged to learn the cue-stimulus associations and pay
487 attention to them. Moreover, a study where cueing was used for pain conditioning reports
488 activations in the periaqueductal gray when contrasting cues anticipating uncertain pain
489 intensity to those anticipating certain pain intensity (101). Critically, (101) also instructed
490 participants to pay attention to the cues. Since attention enhances overall cortical
491 responsiveness as well as predictability effects (102,103), the lack of predictability effects may
492 be a consequence of the lack of such instructions. However, (104) still found predictability
493 effects in ventral visual areas, despite instructing the participants that the contingencies were
494 not task relevant, similarly to our study and to that of (74). Another hypothesis is that, as (105)
495 suggest, predictability effects in cueing paradigms depend on the presence of a neutral
496 condition (i.e., when a cue is uninformative about the occurrence of stimulus type). Whereas
497 (104), (106) and (101) only included expected and unexpected conditions, our study and that
498 by (74) also included a neutral condition. Moreover, a recent study using uni- and multi-variate
499 analytic approaches on EEG data from a large sample during a cueing paradigm with
500 contingencies similar to ours found no effects of predictability (75). Notably, in (75)
501 participants were intensively and successfully trained to learn the cue-stimulus associations, so
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
23
502 the lack of effects cannot be ascribed to lack of attention or failure to use the cues. However,
503 even the idea that the effect can only be found in the absence of a neutral condition is debatable,
504 as (100) report predictability effects face- and object- sensitive regions in the presence of an
505 uninformative condition similar to ours. Similarly, evidence is mixed even regarding whether
506 participants develop predictions about the single stimulus or, rather, about a stimulus category
507 (105). In sum, while it seems to be better to ensure active learning and use of cues, and the
508 effect might be larger in the absence of a neutral condition, cueing designs appear to be less
509 robust as previously thought in producing predictability effects. Ideally, a highly-powered,
510 multi-site and multi-method study should systematically assess the impact of task instructions,
511 attention, cue-stimulus contingencies, and category vs. stimulus learning (see further
512 recommendations in (105)).
513 Another important finding is that, in contrast to what (74) reported, we failed to detect
514 an interaction between predictability and category, effect on which we based our power
515 analysis. This discrepancy between our results and that of (74), using nearly identical
516 paradigms and instructions, suggests that this effect might not necessarily be robust across
517 smaller differences between the studies (including, for instance, the specific stimulus set used).
518 When inspecting responses for typical and distinctive stimuli separately (Figure 3), we
519 observed a pattern resembling the interaction found by (74), but for typical stimuli only.
520 Although the three-way interaction between category, typicality and predictability was not
521 significant, this pattern suggests that the highly typical faces and buildings used by (74) may
522 have produced enhanced category processing despite the lower number of participants in their
523 study.
524 4.3 Future Investigations on the Interaction between Typicality and Predictability
525 Critical to our original research question, we detected no interaction between stimulus
526 typicality and predictability, neither in our ROIs, nor at the whole-brain level. We neither found
527 any three-way interaction between typicality, category, and predictability at the ROI level (see
528 Table 1). Our main explanation for the lack of such interactions is the lack of predictability
529 effects the present study, discussed in Section 4.2. An alternative possibility is that, in our
530 manipulations, typicality is formed through years of exposure to stimuli (corresponding to what
531 (1) called “structural typicality”). In contrast, predictability was only induced by verbal
532 instructions, and via prior learning of cue-stimulus associations bearing no semantic values in
533 themselves, and was not task-relevant. The fact that we found strong effects of typicality
534 whereas predictability effects were absent in the regions of interest and in the rest of the brain
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
24
535 suggests that the two manipulations did not affect neural processing comparably (also see
536 Section 4.3). Finally, we cannot completely exclude the idea that typicality simply does not
537 modulate predictability effects. This would imply a theoretical separation between typicality-
538 and predictability-based neural processing advantages. While predictable stimuli invariably
539 produce smaller prediction error signals (107,108), lower responses to typical stimuli might
540 reflect mechanisms such as neural sharpening (109) or facilitation (110) to commonly seen
541 features.
542 However, at present these results highlight that cueing designs might not lead to highly
543 robust neural effects, and this affects our ability to make conclusive statements on the presence
544 of interactive vs. additive effects between the two variables, and on their respective theoretical
545 implications. As stated above, further research is needed to obtain a robust cueing paradigm.
546 Afterwards, future studies may test whether typicality and predictability rely on similar or
547 related neural mechanisms by using novel stimulus categories (e.g., (24)), another, different
548 predictability-inducing paradigm (see (83) for suggestions) and compare different models of
549 neural responses (74). Such experiments would allow us to i) manipulate predictability and
550 typicality formed at the same time, ii) detect predictability effects that are strong enough to
551 open the possibility for an interaction (e.g., (107,111)), and iii) establish whether the neural
552 mechanisms driving typicality and predictability are the same.
553 4.4 Category Effects
554 Finally, our ROI analysis revealed clear differences in the effects of stimulus category
555 between the FFA and the LOC, with stronger responses to faces and to chairs in the two
556 respective areas. In addition, we observed activations in object-responsive regions in the
557 ventral visual stream for chairs (89,90). By contrast, the whole-brain responses to faces did not
558 reveal any regions of the core face processing network (112,113). Instead, faces elicited
559 stronger responses in the rSTG. Typically, studies report face responses in the right
560 inferior/middle temporal gyri (see convergence results in (114) and (113)) and in the superior
561 temporal sulcus (reviewed by (115)), whereas the STG is rarely reported in face research.
562 However, some studies report increased responses in the rSTG to other people’s faces as
563 compared to one’s own (116), and responses to the opposite contrast in its left homotopic area
564 (117,118). In our task, participants saw faces that looked like people their age, that they could
565 encounter in real life, so they might have recruited self-other distinction processes more in
566 response to faces, as compared to chairs.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
25
567 5. Conclusions
568 We found that distinctive stimuli elicited larger responses in face- and object-
569 responsive regions of the ventral visual stream in ROI analyses, and spatially extended effects
570 encompassing these regions when considering whole-brain analyses. The stronger brain
571 responses to distinctive stimuli are in line with the idea that stimuli that are distant from the
572 prototype (or general tendency of a given category) require more neural resources, and thus
573 produce increased signals. These could be interpreted as a signal of distance from the prototype
574 in the context of prototype-based models, or a rarity signal in the context of exemplar-based
575 models. The present typicality effects seemed to be independent of cue-induced predictability.
576 However, we failed at producing predictability effects with our cueing paradigm and,
577 importantly, we did not replicate the interaction between predictability and category reported
578 by (74). Thus, our research question on the interaction between typicality and predictability
579 remains open, and we provide here some directions for future studies.
580
581 Acknowledgements: The authors gratefully acknowledge Richard Jahn and Daniel Güllmar
582 for their organizational support during data collection, Julian Kauk for his expert support with
583 data analysis, and our experimental participants for contributing their data to this study.
584 Chenglin Li was supported by the China Scholarship Council (CSC) scholarship
585 (201808330399) during the study. Open Access funding enabled and organized by Projekt
586 DEAL.
587
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
26
588
589 6. References
590 1. Dieciuc MA, Folstein JR. Typicality: Stable structures and flexible functions. Psychon
591 Bull Rev [Internet]. 2019;26(2):491–505. Available from:
592 https://doi.org/10.3758/s13423-018-1546-2
593 2. Drucker DM, Aguirre GK. Different Spatial Scales of Shape Similarity Representation
594 in Lateral and Ventral LOC. Cereb Cortex [Internet]. 2009;19(10):2269–80. Available
595 from: https://doi.org/10.1093/cercor/bhn244
596 3. Edelman S. Representation is representation of similarities. Behav Brain Sci. 1998
597 Aug;21(4):449–98.
598 4. Gaißert N, Wallraven C, Bülthoff HH. Visual and haptic perceptual spaces show high
599 similarity in humans. J Vis [Internet]. 2010;10(11):2. Available from:
600 https://doi.org/10.1167/10.11.2
601 5. Rosch E, Mervis CB. Family resemblances: Studies in the internal structure of
602 categories. Cogn Psychol. 1975;7:573–605.
603 6. Valentine T. A unified account of the effects of distinctiveness, inversion, and race in
604 face recognition. Q J Exp Psychol Sect A. 1991;43(2):161–204.
605 7. Valentine T, Bruce V. Recognizing familiar faces: the role of distinctiveness and
606 familiarity. Can J Psychol. 1986 Sep;40(3):300–5.
607 8. Neumann PG. Visual prototype formation with discontinuous representation of
608 dimensions of variability. Vol. 5, Memory & Cognition. US: Psychonomic Society;
609 1977. p. 187–97.
610 9. Tanaka JW, Kantner J, Bartlett M. How category structure influences the perception of
611 object similarity: The atypicality bias. Front Psychol. 2012;3(JUN):1–11.
612 10. Wuttke SJ, Schweinberger SR. The P200 predominantly reflects distance-to-norm in
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
27
613 face space whereas the N250 reflects activation of identity-specific representations of
614 known faces. Biol Psychol [Internet]. 2019;140(November 2018):86–95. Available
615 from: https://doi.org/10.1016/j.biopsycho.2018.11.011
616 11. Valentine T, Lewis MB, Hills PJ. Face-space: A unifying concept in face recognition
617 research. Q J Exp Psychol. 2016;69(10):1996–2019.
618 12. Jenkins R, Dowsett AJ, Burton AM. How many faces do people know? Proc R Soc B
619 Biol Sci. 2018;285(1888).
620 13. Crouzet SM, Kirchner H, Thorpe SJ. Fast saccades toward faces: Face detection in just
621 100 ms. J Vis [Internet]. 2010;10(4):16. Available from:
622 https://doi.org/10.1167/10.4.16
623 14. Ro T, Russell C, Lavie N. Changing Faces: A Detection Advantage in the Flicker
624 Paradigm. Psychol Sci [Internet]. 2001 Jan 1;12(1):94–9. Available from:
625 https://doi.org/10.1111/1467-9280.00317
626 15. Carbon C-C, Strobach T, Langton SRH, Harsányi G, Leder H, Kovács G. Adaptation
627 effects of highly familiar faces: Immediate and long lasting. Mem Cognit [Internet].
628 2007;35(8):1966–76. Available from: https://doi.org/10.3758/BF03192929
629 16. Faerber SJ, Kaufmann JM, Schweinberger SR. Early temporal negativity is sensitive to
630 perceived (rather than physical) facial identity. Neuropsychologia [Internet].
631 2015;75:132–42. Available from:
632 http://www.sciencedirect.com/science/article/pii/S0028393215300385
633 17. Kloth N, Rhodes G, Schweinberger SR. Watching the brain recalibrate: Neural
634 correlates of renormalization during face adaptation. Neuroimage. 2017;155:1–9.
635 18. Rhodes G, Jeffery L. Adaptive norm-based coding of facial identity. Vision Res.
636 2006;46(18):2977–87.
637 19. Faerber SJ, Kaufmann JM, Leder H, Martin EM, Schweinberger SR. The role of
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
28
638 familiarity for representations in norm-based face space. PLoS One. 2016;11(5):1–15.
639 20. Das-Smaal EA, De Swart JH. Variation within categories. Acta Psychol (Amst)
640 [Internet]. 1984;57(3):165–92. Available from:
641 https://www.sciencedirect.com/science/article/pii/0001691884900301
642 21. Ross DA, Deroche M, Palmeri TJ. Not Just the Norm: Exemplar-Based Models also
643 Predict Face Aftereffects. Psychol Bullettin [Internet]. 2014;21(1):47–70. Available
644 from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3624763/pdf/nihms412728.pdf
645 22. Lewis M. FacespaceR: Towards a unified account of face recognition. Vis cogn
646 [Internet]. 2004 Jan 1;11(1):29–69. Available from:
647 https://doi.org/10.1080/13506280344000194
648 23. Giese MA, Leopold DA. Physiologically inspired neural model for the encoding of
649 face spaces. Neurocomputing [Internet]. 2005;65–66:93–101. Available from:
650 https://www.sciencedirect.com/science/article/pii/S0925231204004485
651 24. Bowman CR, Iwashita T, Zeithamova D. Tracking prototype and exemplar
652 representations in the brain across learning. Behrens TE, Barense M, Barense M,
653 Tompary A, editors. Elife [Internet]. 2020 Nov;9:e59360. Available from:
654 https://doi.org/10.7554/eLife.59360
655 25. Panis S, Vangeneugden J, Op de Beeck HP, Wagemans J. The representation of
656 subordinate shape similarity in human occipitotemporal cortex. J Vis. 2008;8(10):1–
657 15.
658 26. Panis S, Wagemans J, Op de Beeck HP. Dynamic norm-based encoding for unfamiliar
659 shapes in human visual cortex. J Cogn Neurosci. 2011;23(7):1829–43.
660 27. Wallis G. Toward a unified model of face and object recognition in the human visual
661 system. Front Psychol. 2013;4(AUG):1–25.
662 28. Carlin JD, Kriegeskorte N. Adjudicating between face-coding models with individual-
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
29
663 face fMRI responses. PLoS Comput Biol. 2017;13(7):1–28.
664 29. Iordan MC, Greene MR, Beck DM, Fei-Fei L. Typicality sharpens category
665 representations in object-selective cortex. Neuroimage. 2016;134:170–9.
666 30. Mack ML, Preston AR, Love BC. Decoding the Brain’s Algorithm for Categorization
667 from Its Neural Implementation. Curr Biol [Internet]. 2013;23(20):2023–7. Available
668 from: https://www.sciencedirect.com/science/article/pii/S0960982213010415
669 31. Beck A-K, Czernochowski D, Lachmann T, Barahona-Correa B, Carmo JC. Is the
670 dolphin a fish? ERP evidence for the impact of typicality during early visual
671 processing in ultra-rapid semantic categorization in autism spectrum disorder. J
672 Neurodev Disord. 2022;14(1):1–20.
673 32. Holmes SJ, Ellis AW. Age of acquisition and typicality effects in three object
674 processing tasks. Vis cogn [Internet]. 2006 May 1;13(7–8):884–910. Available from:
675 https://doi.org/10.1080/13506280544000093
676 33. Rosch E, Simpson C, Miller RS. Structural bases of typicality effects. J Exp Psychol
677 Hum Percept Perform. 1976;2:491–502.
678 34. Valentine T, Bruce V. The effects of distinctiveness in recognising and classifying
679 faces. Perception. 1986;15(5):525–35.
680 35. Kayaert G, Op de Beeck HP, Wagemans J. Dynamic prototypicality effects in visual
681 search. J Exp Psychol Gen. 2011;140(3):506–19.
682 36. Prunty JE, Jenkins R, Qarooni R, Bindemann M. Ingroup and outgroup differences in
683 face detection. Br J Psychol [Internet]. 2023 May 1;114(S1):94–111. Available from:
684 https://doi.org/10.1111/bjop.12588
685 37. Bartlett JC, Hurry S, Thorley W. Typicality and familiarity of faces. Mem \& Cogn.
686 1984;12:219–28.
687 38. Kaufmann JM, Schulz C, Schweinberger SR. High and low performers differ in the
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
30
688 use of shape information for face recognition. Neuropsychologia. 2013;51(7):1310–9.
689 39. Kaufmann J, Vogt S, Schweinberger S. The big nose bias, or when distinctiveness
690 hinders face learning: Evoking an other-race effect with selectively manipulated same-
691 race faces. J Vis. 2018;18(10):1101.
692 40. Lee KJ, Perrett D. Presentation-Time Measures of the Effects of Manipulations in
693 Colour Space on Discrimination of Famous Faces. Perception [Internet]. 1997 Jun
694 1;26(6):733–52. Available from: https://doi.org/10.1068/p260733
695 41. Light LL, Kayra-Stuart F, Hollander S. Recognition memory for typical and unusual
696 faces. J Exp Psychol Hum Learn. 1979 May;5(3):212–28.
697 42. Posner MI, Keele SW. On the Genesis of Abstract Ideas. Cognitive psychology: Key
698 readings. New York, NY, US: Psychology Press; 2004. 472–481 p. (Key readings in
699 cognition.).
700 43. Mauro R, Kubovy M. Caricature and face recognition. Mem Cognit [Internet].
701 1992;20(4):433–40. Available from: https://doi.org/10.3758/BF03210927
702 44. Rhodes G. Superportraits: Caricatures and recognition. Psychology Press; 1997.
703 45. Wallis G, Siebeck UE, Swann K, Blanz V, Bülthoff HH. The prototype effect
704 revisited: Evidence for an abstract feature model of face recognition. J Vis [Internet].
705 2008 Mar 24;8(3):20. Available from: https://doi.org/10.1167/8.3.20
706 46. Schulz C, Kaufmann JM, Walther L, Schweinberger SR. Effects of anticaricaturing vs.
707 caricaturing and their neural correlates elucidate a role of shape for face learning.
708 Neuropsychologia. 2012 Aug;50(10):2426–34.
709 47. Winograd E. Elaboration and distinctiveness in memory for faces. Vol. 7, Journal of
710 Experimental Psychology: Human Learning and Memory. US: American
711 Psychological Association; 1981. p. 181–90.
712 48. Kaufmann JM, Schweinberger SR. Distortions in the brain? ERP effects of
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
31
713 caricaturing familiar and unfamiliar faces. Brain Res. 2008;1228:177–88.
714 49. Itz ML, Golle J, Luttmann S, Schweinberger SR, Kaufmann JM. Dominance of texture
715 over shape in facial identity processing is modulated by individual abilities. Br J
716 Psychol. 2017;108(2):369–96.
717 50. O’Reilly RC, Rudy JW. Conjunctive representations in learning and memory:
718 principles of cortical and hippocampal function. Psychol Rev. 2001 Apr;108(2):311–
719 45.
720 51. Davis T, Poldrack RA. Measuring neural representations with fMRI: practices and
721 pitfalls. Ann N Y Acad Sci [Internet]. 2013;1296(1):108–34. Available from:
722 https://nyaspubs.onlinelibrary.wiley.com/doi/abs/10.1111/nyas.12156
723 52. Davis T, Poldrack RA. Quantifying the Internal Structure of Categories Using a Neural
724 Typicality Measure. Cereb Cortex [Internet]. 2013;24(7):1720–37. Available from:
725 https://doi.org/10.1093/cercor/bht014
726 53. Loffler G, Yourganov G, Wilkinson F, Wilson HR. fMRI evidence for the neural
727 representation of faces. Nat Neurosci. 2005;8(10):1386–91.
728 54. Bestelmeyer PEG, Latinus M, Bruckert L, Rouger J, Crabbe F, Belin P. Implicitly
729 perceived vocal attractiveness modulates prefrontal cortex activity. Cereb Cortex.
730 2012;22(6):1263–70.
731 55. Latinus M, McAleer P, Bestelmeyer PEG, Belin P. Norm-Based Coding of Voice
732 Identity in Human Auditory Cortex. Curr Biol [Internet]. 2013;23(12):1075–80.
733 Available from:
734 https://www.sciencedirect.com/science/article/pii/S096098221300496X
735 56. Halit H, De Haan M, Johnson MH. Modulation of event-related potentials by
736 prototypical and atypical faces. Neuroreport. 2000;11(9):1871–5.
737 57. Kaufmann JM, Schweinberger SR. The faces you remember: Caricaturing shape
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
32
738 facilitates brain processes reflecting the acquisition of new face representations. Biol
739 Psychol. 2012;89(1):21–33.
740 58. Schröger A, Ficco L, Wuttke SJ, Kaufmann JM, Schweinberger SR. Differences
741 between high and low performers in face recognition in electrophysiological correlates
742 of face familiarity and distance-to-norm. 2023.
743 59. Hauk O, Patterson K, Woollams A, Cooper-Pye E, Pulvermüller F, Rogers TT. How
744 the camel lost its hump: the impact of object typicality on event-related potential
745 signals in object decision. J Cogn Neurosci. 2007 Aug;19(8):1338–53.
746 60. Ellis AE, Nelson CA. Category prototypicality judgments in adults and children:
747 Behavioral and electrophysiological correlates. Dev Neuropsychol. 1999;15(2):193–
748 211.
749 61. Dunn BR, Dunn DA, Languis M, Andrews D. The Relation of ERP Components to
750 Complex Memory Processing. Brain Cogn [Internet]. 1998;36(3):355–76. Available
751 from: https://www.sciencedirect.com/science/article/pii/S0278262698909988
752 62. Rao RP, Ballard DH. Predictive coding in the visual cortex: a functional interpretation
753 of some extra-classical receptive-field effects. Nat Neurosci. 1999 Jan;2(1):79–87.
754 63. Clark A. Whatever next? Predictive brains, situated agents, and the future of cognitive
755 science. Behav Brain Sci [Internet]. 2013;36(3):181–204. Available from:
756 http://www.ncbi.nlm.nih.gov/pubmed/23663408
757 64. Friston K. The free-energy principle: a unified brain theory? Nat Rev Neurosci
758 [Internet]. 2010;11(2):127–38. Available from:
759 http://www.ncbi.nlm.nih.gov/pubmed/20068583
760 65. Friston K, Kiebel S. Predictive coding under the free-energy principle. Philos Trans R
761 Soc B Biol Sci. 2009;364(1521):1211–21.
762 66. Bastos AM, Usrey WM, Adams RA, Mangun GR, Fries P, Friston KJ. Canonical
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
33
763 Microcircuits for Predictive Coding. Neuron [Internet]. 2012;76(4):695–711. Available
764 from: http://dx.doi.org/10.1016/j.neuron.2012.10.038
765 67. Keller GB, Mrsic-Flogel TD. Predictive Processing: A Canonical Cortical
766 Computation. Neuron [Internet]. 2018;100(2):424–35. Available from:
767 https://doi.org/10.1016/j.neuron.2018.10.003
768 68. D’Astolfo L, Rief W. Learning about expectation violation from prediction error
769 paradigms - A meta-analysis on brain processes following a prediction error. Front
770 Psychol. 2017;8(JUL):1–11.
771 69. Hohwy J. Prediction error minimization in the brain. Routledge Handb Comput mind.
772 2018;159–72.
773 70. Friston K. The history of the future of the Bayesian brain. Neuroimage [Internet].
774 2012;62(2):1230–3. Available from:
775 https://www.sciencedirect.com/science/article/pii/S1053811911011657
776 71. Knill DC, Pouget A. The Bayesian brain: the role of uncertainty in neural coding and
777 computation. Trends Neurosci [Internet]. 2004;27(12):712–9. Available from:
778 https://www.sciencedirect.com/science/article/pii/S0166223604003352
779 72. Ganel T, Gonzalez CLR, Valyear KF, Culham JC, Goodale MA, Köhler S. The
780 relationship between fMRI adaptation and repetition priming. Neuroimage [Internet].
781 2006;32(3):1432–40. Available from:
782 https://www.sciencedirect.com/science/article/pii/S1053811906006318
783 73. S. S. Separate modifiability, mental modules, and the use of pure and composite
784 measures to reveal them. Acta Psychol (Amst) [Internet]. 2001;106(1–2):147–246.
785 Available from:
786 http://www.embase.com/search/results?subaction=viewrecord&from=export&id=L334
787 54553
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
34
788 74. Egner T, Monti JM, Summerfield C. Expectation and surprise determine neural
789 population responses in the ventral visual stream. J Neurosci. 2010;30(49):16601–8.
790 75. den Ouden C, Zhou A, Mepani V, Kovács G, Vogels R, Feuerriegel D. Stimulus
791 expectations do not modulate visual event-related potentials in probabilistic cueing
792 designs. Neuroimage. 2023;280(April).
793 76. Lakens D, Caldwell AR. Simulation-Based Power Analysis for Factorial Analysis of
794 Variance Designs. Adv Methods Pract Psychol Sci [Internet]. 2021 Jan
795 1;4(1):2515245920951503. Available from:
796 https://doi.org/10.1177/2515245920951503
797 77. Dennett H, Edwards M, McKone E. Are objects like faces? Norm-based versus
798 exemplar-based coding as revealed by adaptation aftereffects. J Vis [Internet]. 2009
799 Aug 1;9(8):518. Available from: https://doi.org/10.1167/9.8.518
800 78. Brainard DH. The Psychophysics Toolbox. Spat Vis. 1997;10(4):433–6.
801 79. Kleiner M, Brainard DH, Pelli D. What’s new in Psychtoolbox-3? Perception.
802 2007;36:1–16.
803 80. Itz ML, Schweinberger SR, Schulz C, Kaufmann JM. Neural correlates of facilitations
804 in face learning by selective caricaturing of facial shape or reflectance. Neuroimage
805 [Internet]. 2014;102:736–47. Available from:
806 http://www.sciencedirect.com/science/article/pii/S1053811914007162
807 81. Limbach K, Itz ML, Schweinberger SR, Jentsch AD, Romanova L, Kaufmann JM.
808 Neurocognitive effects of a training program for poor face recognizers using shape and
809 texture caricatures: A pilot investigation. Neuropsychologia. 2022;165:1–16.
810 82. Zhou X, Itz ML, Vogt S, Kaufmann JM, Schweinberger SR, Mondloch CJ. Similar use
811 of shape and texture cues for own- and other-race faces during face learning and
812 recognition. Vision Res [Internet]. 2021;188(September 2019):32–41. Available from:
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
35
813 https://doi.org/10.1016/j.visres.2021.06.014
814 83. Feuerriegel D. Selecting appropriate designs and comparison conditions in repetition
815 paradigms. Cortex [Internet]. 2016;80:196–205. Available from:
816 http://www.sciencedirect.com/science/article/pii/S0010945215003743
817 84. Cziraki C, Greenlee MW, Kovács G. Neural Correlates of High-Level Adaptation-
818 Related Aftereffects. J Neurophysiol [Internet]. 2010 Jan 13;103(3):1410–7. Available
819 from: https://doi.org/10.1152/jn.00582.2009
820 85. Caspers J, Zilles K, Eickhoff SB, Schleicher A, Mohlberg H, Amunts K.
821 Cytoarchitectonical analysis and probabilistic mapping of two extrastriate areas of the
822 human posterior fusiform gyrus. Brain Struct Funct [Internet]. 2013;218(2):511–26.
823 Available from: https://doi.org/10.1007/s00429-012-0411-8
824 86. Kanwisher N, McDermott J, Chun MM. The Fusiform Face Area: A Module in Human
825 Extrastriate Cortex Specialized for Face Perception. J Neurosci [Internet]. 1997 Jun
826 1;17(11):4302 LP – 4311. Available from:
827 http://www.jneurosci.org/content/17/11/4302.abstract
828 87. Grill-Spector K, Kushnir T, Hendler T, Edelman S, Itzchak Y, Malach R. A sequence
829 of object-processing stages revealed by fMRI in the human occipital lobe. Hum Brain
830 Mapp. 1998;6(4):316–28.
831 88. Malikovic A, Amunts K, Schleicher A, Mohlberg H, Kujovic M, Palomero-Gallagher
832 N, et al. Cytoarchitecture of the human lateral occipital cortex: mapping of two
833 extrastriate areas hOc4la and hOc4lp. Brain Struct Funct [Internet]. 2016;221(4):1877–
834 97. Available from: https://doi.org/10.1007/s00429-015-1009-8
835 89. Grill-Spector K, Kourtzi Z, Kanwisher N. The lateral occipital complex and its role in
836 object recognition. Vision Res [Internet]. 2001;41(10):1409–22. Available from:
837 https://www.sciencedirect.com/science/article/pii/S0042698901000736
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
36
838 90. Malach R, Reppas JB, Benson RR, Kwong KK, Jiang H, Kennedy WA, et al. Object-
839 related activity revealed by functional magnetic resonance imaging in human occipital
840 cortex. Proc Natl Acad Sci. 1995;92(18):8135–9.
841 91. Ollinger JM, Shulman GL, Corbetta M. Separating Processes within a Trial in Event-
842 Related Functional MRI: I. The Method. Neuroimage [Internet]. 2001;13(1):210–7.
843 Available from:
844 https://www.sciencedirect.com/science/article/pii/S1053811900907109
845 92. Brett M, Anton J-L, Valabregue R, Poline J-B. Region of interest analysis using an
846 SPM toolbox. In: 8th international conference on functional mapping of the human
847 brain. Sendai; 2002. p. 497.
848 93. Eickhoff SB, Stephan KE, Mohlberg H, Grefkes C, Fink GR, Amunts K, et al. A new
849 SPM toolbox for combining probabilistic cytoarchitectonic maps and functional
850 imaging data. Neuroimage [Internet]. 2005;25(4):1325–35. Available from:
851 https://www.sciencedirect.com/science/article/pii/S105381190400792X
852 94. Lancaster JL, Woldorff MG, Parsons LM, Liotti M, Freitas CS, Rainey L, et al.
853 Automated Talairach Atlas labels for functional brain mapping. Hum Brain Mapp.
854 2000;10(3):120–31.
855 95. Lancaster JL, Rainey LH, Summerlin JL, Freitas CS, Fox PT, Evans AC, et al.
856 Automated labeling of the human brain: A preliminary report on the development and
857 evaluation of a forward-transform method. Hum Brain Mapp. 1997;5(4):238–42.
858 96. van Doorn J, van den Bergh D, Böhm U, Dablander F, Derks K, Draws T, et al. The
859 JASP guidelines for conducting and reporting a Bayesian analysis. Psychon Bull Rev.
860 2021 Jun;28(3):813–26.
861 97. Ashby FG, Maddox WT. Relations between prototype, exemplar, and decision bound
862 models of categorization. J Math Psychol. 1993;37(3):372–400.
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
37
863 98. Sigala N, Gabbiani F, Logothetis NK. Visual Categorization and Object
864 Representation in Monkeys and Humans. J Cogn Neurosci [Internet]. 2002 Feb
865 15;14(2):187–98. Available from: https://doi.org/10.1162/089892902317236830
866 99. Grotheer M, Kovacs GZ. Can predictive coding explain repetition suppression? Cortex
867 [Internet]. 2016;80(Review):113–24. Available from:
868 https://www.sciencedirect.com/science/article/pii/S0010945216000149
869 100. Dunovan K, Wheeler ME. Computational and neural signatures of pre and post-
870 sensory expectation bias in inferior temporal cortex. Sci Rep [Internet].
871 2018;8(1):13256. Available from: https://doi.org/10.1038/s41598-018-31678-x
872 101. Lin C-S, Hsieh J-C, Yeh T-C, Niddam DM. Predictability-mediated pain modulation
873 in context of multiple cues: An event-related fMRI study. Neuropsychologia [Internet].
874 2014;64:85–91. Available from:
875 https://www.sciencedirect.com/science/article/pii/S0028393214003261
876 102. Kok P, Rahnev D, Jehee JFM, Lau HC, De Lange FP. Attention reverses the effect of
877 prediction in silencing sensory signals. Cereb Cortex. 2012;22(9):2197–206.
878 103. Hsu Y-F, Hamalainen J, Waszak F. Both attention and prediction are necessary for
879 adaptive neuronal tuning in sensory processing. Front Hum Neurosci. 2014;8:152.
880 104. Richter D, de Lange FP. Statistical learning attenuates visual activity only for attended
881 stimuli. Frank MJ, Kahnt T, Wyart V, Heinzle J, editors. Elife [Internet].
882 2019;8:e47869. Available from: https://doi.org/10.7554/eLife.47869
883 105. Feuerriegel D, Vogels R, Kovács G. Evaluating the evidence for expectation
884 suppression in the visual system. Neurosci Biobehav Rev. 2021;126(February):368–
885 81.
886 106. Davis B, Hasson U. Predictability of what or where reduces brain activity , but a
887 bottleneck occurs when both are predictable . Neuroimage [Internet]. 2018;167(May
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
38
888 2016):224–36. Available from: https://doi.org/10.1016/j.neuroimage.2016.06.001
889 107. Summerfield C, Monti JM, Trittschuh EH, Mesulam M, Egner T. Neural repetition
890 suppression reflects fulfilled perceptual expectations. Nat Neurosci [Internet].
891 2008;11(9):1004–6. Available from: http://www.ncbi.nlm.nih.gov/pubmed/21138698
892 108. Robinson JE, Woods W, Leung S, Kaufman J, Breakspear M, Young AW, et al.
893 Prediction-error signals to violated expectations about person identity and head
894 orientation are doubly-dissociated across dorsal and ventral visual stream regions.
895 Neuroimage [Internet]. 2020;206(November 2019):116325. Available from:
896 https://doi.org/10.1016/j.neuroimage.2019.116325
897 109. Desimone R. Neural mechanisms for visual memory and their role in attention. Proc
898 Natl Acad Sci U S A. 1996 Nov;93(24):13494–9.
899 110. Henson RNA, Rugg MD. Neural response suppression, haemodynamic repetition
900 effects, and behavioural priming. Neuropsychologia [Internet]. 2003;41(3):263–70.
901 Available from:
902 https://www.sciencedirect.com/science/article/pii/S0028393202001598
903 111. Kovács G, Iffland L, Vidnyánszky Z, Greenlee MW. Stimulus repetition probability
904 effects on repetition suppression are position invariant for faces. Neuroimage.
905 2012;60(4):2128–35.
906 112. Haxby J V, Hoffman EA, Gobbini MI. The distributed human neural system for face
907 perception. Trends Cogn Sci [Internet]. 2000;4(6):223–33. Available from:
908 http://www.sciencedirect.com/science/article/pii/S1364661300014820
909 113. Müller VI, Höhner Y, Eickhoff SB. Influence of task instructions and stimuli on the
910 neural network of face processing: An ALE meta-analysis. Cortex. 2018;103:240–55.
911 114. Fusar-Poli P, Placentino A, Carletti F, Landi P, Allen P, Surguladze S, et al. Functional
912 atlas of emotional faces processing: A voxel-based meta-analysis of 105 functional
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
39
913 magnetic resonance imaging studies. J Psychiatry Neurosci. 2009;34(6):418–32.
914 115. Iidaka T. Role of the fusiform gyrus and superior temporal sulcus in face perception
915 and recognition: An empirical review. Jpn Psychol Res [Internet]. 2014 Jan
916 1;56(1):33–45. Available from: https://doi.org/10.1111/jpr.12018
917 116. Devue C, Collette F, Balteau E, Degueldre C, Luxen A, Maquet P, et al. Here I am:
918 The cortical correlates of visual self-recognition. Vol. 1143, Brain Research. Devue,
919 Christel: Department of Cognitive Science, University of Liege, 5, Boulevard du
920 Rectorat (Bat. B32), Liege, Belgium, 4000, cdevue@ulg.ac.be: Elsevier Science; 2007.
921 p. 169–82.
922 117. Kircher TTJ, Senior C, Phillips ML, Rabe-Hesketh S, Benson PJ, Bullmore ET, et al.
923 Recognizing one’s own face. Cognition [Internet]. 2001;78(1):B1–15. Available from:
924 https://www.sciencedirect.com/science/article/pii/S0010027700001049
925 118. Uddin LQ, Kaplan JT, Molnar-Szakacs I, Zaidel E, Iacoboni M. Self-face recognition
926 activates a frontoparietal “mirror” network in the right hemisphere: an event-related
927 fMRI study. Neuroimage [Internet]. 2005;25(3):926–35. Available from:
928 https://www.sciencedirect.com/science/article/pii/S1053811904007669
929
930
931 Supplementary captions
932
933 Table S1. Summary of post-experimental survey. Synopsis of the post-experimental survey
934 completed by participants Each participant was first asked about whether they experienced any
935 form of discomfort during the procedure, and eventually which ones. Then they were asked to
936 indicate to which degree they found themselves paying attention to cue-category contingencies
937 during the scanning session. Their verbal reports (e.g., “never”, not really”, “at the beginning
938 only”, “sometimes”, etc.) were then transferred to a scale ranging from 0 (= “not at all”) to 3
939 (= “very frequently/always”). No participant reported a score of 3. Participants were then asked
940 whether the faces and the chairs shown during the tasks look ordinary to them or whether they
941 had something distinctive. Their responses were coded as “yes” if they made observations
942 about differential typicality within that category (e.g., “some of the chairs were quite special”),
943 and with “no” if all items of that category appeared of equal typicality to them (e.g., “all faces
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Investigating the neural effects of typicality and predictability for face and object
stimuli
40
944 looked pretty normal to me, none stood out”). As shown in the total proportions of “yes”
945 responses, typicality manipulations were detected much more often in chairs than faces.
946
947 Table S2. ROI locations in individual participants. MNI coordinates of activation peaks in
948 individual participants during functional localizer run. Right and left fusiform faces areas
949 (FFAs) were localized using the contrast Faces > Objects (or Faces > Noise when the first did
950 not reveal any peak in the approximate area). Right and left lateral occipital complexes (LOCs)
951 were localized using the contrast Objects > Noise. Stars (*) indicate that the coordinate was
952 retrieved when changing the threshold from p_FWE<.05 to p_uncorrected<.001. In a few cases
953 no peaks were found for a particular area. When comparing this table to Fig. S1 and Fig. 3,
954 note that even though we were able to locate these ROIs in nearly all participants, at times the
955 hemodynamic function extracted at the respective location was not of sufficient quality, thus
956 data were excluded from the analysis reported in the main text.
957
958 Table S3. Contrast weights for the whole-brain analysis. Conditions are abbreviated as
959 follows: H = high predictability cue (75% contingencies); M = medium/uninformative cue
960 (50% contingencies); L = low predictability cue (25 % contingencies); F = face stimulus; C =
961 chair stimulus; T = typical stimulus; D = distinctive stimulus. Note that weights are set to 0 for
962 target trials and for the six nuisance regressors.
963
964 Figure S1. ANOVA results in individual ROIs.
965
966 Figure S2: Whole-brain typicality effects for faces and chairs separately. Whole-brain
967 effects of typicality in faces and chairs separately (directional t-contrasts). Results for the
968 contrast distinctive chairs > typical chairs are thresholded at an alpha level of .05 and a family-
969 wise error correction for multiple comparison (FWE, p < .05). The other contrasts are
970 thresholded at an alpha level of .001 and no correction (uncorrected, p < .001). The ink
971 transparency in the legend visually conveys the different thresholding conservativeness of the
972 results. This figure shows that the effects of distinctiveness (distinctive > typical) tend to occur
973 in overlapping regions for both stimulus types (note that, with an uncorrected threshold, the
974 clusters related to chair distinctiveness overlap with those related to face distinctiveness).
975 Conversely, typicality effects seem to be differently distributed and to be more prominent for
976 chairs.
977
978
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Face stimuli are not shown
due to preprint policy due to preprint policy
Face stimuli are not shown
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
Face stimuli not shown
due to preprint policy
Face stimuli not shown
due to preprint policy
Face stimuli not shown
due to preprint policy
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
.CC-BY 4.0 International licenseperpetuity. It is made available under a
preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in
The copyright holder for thisthis version posted October 27, 2023. ; https://doi.org/10.1101/2023.10.24.563242doi: bioRxiv preprint
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Background Neurotypical individuals categorize items even during ultra-rapid presentations (20 ms; see Thorpe et al. Nature 381: 520, 1996). In cognitively able autistic adults, these semantic categorization processes may be impaired and/or may require additional time, specifically for the categorization of atypical compared to typical items. Here, we investigated how typicality structures influence ultra-rapid categorization in cognitively able autistic and neurotypical male adults. Methods Images representing typical or atypical exemplars of two different categories (food/animals) were presented for 23.5 vs. 82.3 ms (short/long). We analyzed detection rates, reaction times, and the event-related potential components dN150, N1, P2, N2, and P3 for each group. Results Behavioral results suggest slower and less correct responses to atypical compared to typical images. This typicality effect was larger for the category with less distinct boundaries (food) and observed in both groups. However, electrophysiological data indicate a different time course of typicality effects, suggesting that neurotypical adults categorize atypical images based on simple features (P2), whereas cognitively able autistic adults categorize later, based on arbitrary features of atypical images (P3). Conclusions We found evidence that all three factors under investigation — category, typicality, and presentation time — modulated specific aspects of semantic categorization. Additionally, we observed a qualitatively different pattern in the autistic adults, which suggests that they relied on different cognitive processes to complete the task.
Article
Full-text available
Humans show improved recognition for faces from their own social group relative to faces from another social group. Yet before faces can be recognised, they must first be detected in the visual field. Here, we tested whether humans also show an ingroup bias at the earliest stage of face processing – the point at which the presence of a face is first detected. To this end, we measured viewers’ ability to detect ingroup (Black, White) and outgroup faces (Asian, Black, White) in everyday scenes. Ingroup faces were detected with greater speed and accuracy relative to outgroup faces (Experiment 1). Removing face hue impaired detection generally, but the ingroup detection advantage was undiminished (Experiment 2). This same pattern was replicated by a detection algorithm using face templates derived from human data (Experiment 3). These findings demonstrate that the established ingroup bias in face processing can extend to the early process of detection. This effect is ‘colour blind’, in the sense that group membership effects are independent of general effects of image hue. Moreover, it can be captured by tuning visual templates to reflect the statistics of observers’ social experience. We conclude that group bias in face detection is both a visual and a social phenomenon.
Article
Full-text available
Researchers often rely on analysis of variance (ANOVA) when they report results of experiments. To ensure that a study is adequately powered to yield informative results with an ANOVA, researchers can perform an a priori power analysis. However, power analysis for factorial ANOVA designs is often a challenge. Current software solutions do not allow power analyses for complex designs with several within-participants factors. Moreover, power analyses often need [Formula: see text] or Cohen’s f as input, but these effect sizes are not intuitive and do not generalize to different experimental designs. We have created the R package Superpower and online Shiny apps to enable researchers without extensive programming experience to perform simulation-based power analysis for ANOVA designs of up to three within- or between-participants factors. Predicted effects are entered by specifying means, standard deviations, and, for within-participants factors, the correlations. The simulation provides the statistical power for all ANOVA main effects, interactions, and individual comparisons. The software can plot power across a range of sample sizes, can control for multiple comparisons, and can compute power when the homogeneity or sphericity assumption is violated. This Tutorial demonstrates how to perform a priori power analysis to design informative studies for main effects, interactions, and individual comparisons and highlights important factors that determine the statistical power for factorial ANOVA designs.
Article
Full-text available
There is a long-standing debate about whether categories are represented by individual category members (exemplars) or by the central tendency abstracted from individual members (prototypes). Neuroimaging studies have shown neural evidence for either exemplar representations or prototype representations, but not both. Presently, we asked whether it is possible for multiple types of category representations to exist within a single task. We designed a categorization task to promote both exemplar and prototype representations and tracked their formation across learning. We found only prototype correlates during the final test. However, interim tests interspersed throughout learning showed prototype and exemplar representations across distinct brain regions that aligned with previous studies: prototypes in ventromedial prefrontal cortex and anterior hippocampus and exemplars in inferior frontal gyrus and lateral parietal cortex. These findings indicate that, under the right circumstances, individuals may form representations at multiple levels of specificity, potentially facilitating a broad range of future decisions.
Article
Full-text available
Over our species history, humans have typically lived in small groups of under a hundred individuals. However, our face recognition abilities appear to equip us to recognize very many individuals, perhaps thousands. Modern society provides access to huge numbers of faces, but no one has established how many faces people actually know. Here, we describe a method for estimating this number. By combining separate measures of recall and recognition, we show that people know about 5000 faces on average and that individual differences are large. Our findings offer a possible explanation for large variation in identification performance. They also provide constraints on understanding the qualitative differences between perception of familiar and unfamiliar faces—a distinction that underlies all current theories of face recognition.
Article
Humans and other animals can learn and exploit repeating patterns that occur within their environments. These learned patterns can be used to form expectations about future sensory events. Several influential predictive coding models have been proposed to explain how learned expectations influence the activity of stimulus-selective neurons in the visual system. These models specify reductions in neural response measures when expectations are fulfilled (termed expectation suppression) and increases following surprising sensory events. However, there is currently scant evidence for expectation suppression in the visual system when confounding factors are taken into account. Effects of surprise have been observed in blood oxygen level dependent (BOLD) signals, but not when using electrophysiological measures. To provide a strong test for expectation suppression and surprise effects we performed a predictive cueing experiment while recording electroencephalographic (EEG) data. Participants (n=48) learned cue-face associations during a training session and were then exposed to these cue-face pairs in a subsequent experiment. Using univariate analyses of face-evoked event-related potentials (ERPs) we did not observe any differences across expected (90% probability), neutral (50%) and surprising (10%) face conditions. Across these comparisons, Bayes factors consistently favoured the null hypothesis throughout the time-course of the stimulus-evoked response. When using multivariate pattern analysis we did not observe above-chance classification of expected and surprising face-evoked ERPs. By contrast, we found robust within- and across-trial stimulus repetition effects. Our findings do not support predictive coding-based accounts that specify reduced prediction error signalling when perceptual expectations are fulfilled. They instead highlight the utility of other types of predictive processing models that describe expectation-related phenomena in the visual system without recourse to prediction error signalling.
Article
Valentine's influential norm-based multidimensional face-space model (nMDFS) predicts that perceived distinctiveness increases with distance to the norm. Occipito-temporal event-related potentials (ERPs) have been recently shown to respond selectively to variations in distance-to-norm (P200) or familiarity (N250, late negativity), respectively (Wuttke & Schweinberger, 2019). Despite growing evidence on interindividual differences in face perception skills at the behavioral level, little research has focused on their electrophysiological correlates. To reveal potential interindividual differences in face spaces, we contrasted high and low performers in face recognition in regards to distance-to-norm (P200) and familiarity (N250). We replicated both the P200 distance-to-norm and the N250 familiarity effect. Importantly, we observed: i) reduced responses in low compared to high performers of face recognition, especially in terms of smaller distance-to-norm effects in the P200, possibly indicating less 'expanded' face spaces in low compared to high performers; ii) increased N250 responses to familiar original faces in high performers, suggesting more robust face identity representations. In summary, these findings suggest the contribution of both early norm-based face coding and robust face representations to individual face recognition skills, and indicate that ERPs can offer a promising route to understand individual differences in face perception and their neurocognitive correlates.
Article
Although the other-race effect (ORE; superior recognition of own- relative to other-race faces) is well established, the mechanisms underlying it are not well understood. We examined whether the ORE is attributable to differential use of shape and texture cues for own- vs. other-race faces. Shape cues are particularly important for detecting that an own-race face is unfamiliar, whereas texture cues are more important for recognizing familiar and newly learned own-race faces. We compared the influence of shape and texture cues on Caucasian participants’ recognition of Caucasian and East Asian faces using two complementary approaches. In Experiment 1, participants studied veridical, shape-caricatured, or texture-caricatured faces and then were asked to recognize them in an old/new recognition task. In Experiment 2, all study faces were veridical and we independently removed the diagnosticity of shape (or texture) cues in the test phase by replacing original shape (or texture) with average shape (or texture). Despite an overall own-race advantage, participants’ use of shape and texture cues was comparable for own- and other-race faces. These results suggest that the other-race effect is not attributable to qualitative differences in the use of shape and texture cues.
Article
Reports of expectation suppression have shaped the development of influential predictive coding-based theories of visual perception. However recent work has highlighted confounding factors that may mimic or inflate expectation suppression effects. In this review, we describe four confounds that are prevalent across experiments that tested for expectation suppression: effects of surprise, attention, stimulus repetition and adaptation, and stimulus novelty. With these confounds in mind we then critically review the evidence for expectation suppression across probabilistic cueing, statistical learning, oddball, action-outcome learning and apparent motion designs. We found evidence for expectation suppression within a specific subset of statistical learning designs that involved weeks of sequence learning prior to neural activity measurement. Across other experimental contexts, whereby stimulus appearance probabilities were learned within one or two testing sessions, there was inconsistent evidence for genuine expectation suppression. We discuss how an absence of expectation suppression could inform models of predictive processing, repetition suppression and perceptual decision-making. We also provide suggestions for designing experiments that may better test for expectation suppression in future work.
Article
This perspective describes predictive processing as a computational framework for understanding cortical function in the context of emerging evidence, with a focus on sensory processing. We discuss how the predictive processing framework may be implemented at the level of cortical circuits and how its implementation could be falsified experimentally. Lastly, we summarize the general implications of predictive processing on cortical function in healthy and diseased states. In this perspective, Keller and Mrsic-Flogel describe the advantages of predictive processing as a computational framework for understanding cortical function in the context of emerging evidence with a focus on sensory processing.