top of page


The group addresses questions, such as what’s the difference between a posed and a spontaneous expression? how fast can we perceive a face or an emotional expression?, what strategies do radiologists employ to detect breast cancer and is this skill trainable? how do clinical conditions, such as depression, autism, affect face recognition? To address these questions, researchers in the collaborative employ a variety of empirical techniques involving psychophysics, cognitive experiments, eye tracking, neural imaging (fMRI, EEG), and computer modeling.  
June 8th, 2022

Jake Martin

High speed continuous visual search in Autism Spectrum Disorder and Schizophrenia.


Schizophrenia and Autism Spectrum Disorder have been proposed and shown to arise from many sorts of putative neurological origins.  Many visual and cognitive models suggest that the circuits that are involved with the initiation and the later processing of the feedback pass can explain parts of the sequelae involved with either disorder.  However, over the past 20 years, it has been shown that the initial feedforward processing is also affected [1].  In this talk, I will present preliminary work wherein 12 neurotypical comparison subjects, 13 SZ, and 6 high-functioning ASD patients did a feedforward-intensive high speed continuous visual search task with only their eyes [2].  In blocks of 500 trials, the participants continuously targeted either small upright faces, inverted faces, or gabor patches of various orientations.  Within each block, the objects were either pasted into a cluttered scene or on a blank screen.  Despite the easy nature of the task, our preliminary analyses indicated striking differences in saccadic eye movement and search performance properties between ASD patients, SZ patients, and neurotypical comparison subjects. These findings may help to confirm and clarify how and where early feedforward visual activity differs between and within these disparate clinical populations.

Link to Zoom Recording of Jake's Talk
March 9th, 2022

Ipek Oruc 

Invariance in high-level visual recognition and its relation to the visual environment


The term ‘invariance’ in visual form recognition refers to a fundamental requirement from any pattern recognition process, whether biological or artificial, that performance should be impervious to image variations that do not alter the identity/category/label of the visual form. For example, in template matching, the system might store a prototypical copy of each pattern (the template) with standard orientation, size, and position. Incoming exemplars, which might be, e.g., crooked, small, and off-center in view, may be snapped back into an upright orientation, resized to standard scale and shifted to the center, before being compared to the templates. This process, thus ensures orientation, scale, and translation invariance. How is invariant recognition accomplished in the visual system? In this talk I will focus on scale-invariance and overview behavioral data we have collected over the years that demonstrate some limits to scale invariance in face and letter recognition. I will also present data from our naturalistic observation study showing statistics of the visual environment that pertain to faces. Together, these data suggest invariant perception in the visual system may be created by patching together piece-meal non-invariant processes tailor-built  for the demands of the visual environment.  

Link to Zoom Recording of Ipek’s Talk

Suggested Readings: 

Oruc, I., & Barton, J. J. (2010). Critical frequencies in the perception of letters, faces, and novel shapes: Evidence for limited scale invariance for faces. Journal of Vision, 10(12), 20-20.
Oruc, I., Shafai, F., Murthy, S., Lages, P., & Ton, T. (2019). The adult face-diet: A naturalistic observation study. Vision research, 157, 222-229.
February 9th, 2022 

Ben Balas 

What do you do with a social face space? A miscellany of wayward results in impression formation.



One thing I learned from watching too many commercials in the '80's is that "You never get a second chance to make a first impression." While I'm not sure vision science has a ton to say yet about those second chances, there is a large literature describing how individuals form impressions of others based on facial appearance. A wide range of social attributes can be reliably (but often not accurately!) estimated from face images, and low-dimensional models of social face space remain a popular way to account for observers' inferences. Our lab has published a few results in this domain, but what I'll talk about here is a collection of results that we *didn't* publish. Each of these is a (possibly misguided) attempt to address some issues that we thought were important to bring up with regard to impression formation, but that haven't made it out the door yet: How invariant is impression formation to image variability? Can you really use a low-dimensional face space to make inferences about lots of different social attributes? How do you get a social face space from images in the first place? In each case, I'll present some results that my students and I think suggest some interesting answers to these questions, but maybe still leave a little too much room for alternate accounts. That is to say that there are almost certainly a lot of flaws in all of these studies, but I think there might still be a few good ideas in there too. By presenting these not-quite-ready-for-primetime results, I hope I can generate a lot of discussion about impression formation research more broadly and maybe spark some ideas for collaborations.

Link to Zoom Recording of Ben’s Talk


Suggested Readings:

Balas, B., & Thrash, J. (2019). Using social face-space to estimate social variables in real and artificial faces.


January 12th, 2022

Karla Evans &  Alan Baddeley 

A Two-Process Account of Long-Term Memory for Visual Scenes 


The series of experiments I present are part of an investigation into the role of attention in visual long-term recognition memory (LTM). The questions addressed are whether longer encoding time leads to a higher level of overall performance and if this is effect differs across different visual materials expressed either in hits or false alarms or both?  In addition, what are the effect of varying the richness of features within the sets of material?  The results show a consistent linear increase in encoding with exposure time in the conventional measure of recognition performance, d’ but for long term visual memory this pattern does not tell the major story. Examining effects of attention on LTM in a more graded way show results that are consistent with a positive role of the amount of attentional capacity almost entirely driven by increase in hits with time while the effects of feature overlap in images or similarity is carried entirely by false alarms. This pattern of results fits very well with a two-stage model which is proposes, an initial extraction of gist minimally influenced by details followed by a more parallel judgment based on detail allowing for differentiation.

Link to Zoom Recording of Karla’s Talk 

Suggested Readings: 

Evans, K. K., & Baddeley, A. (2018). Intention, attention and long-term memory for visual scenes: It all depends on the scenes. Cognition, 180, 24-37

December 8th, 2021

Quoc Vuong 

Perceptual cues to emotions: A developmental approach



Adults can quickly and efficiently extract perceptual cues from body, facial and vocal expressions to recognize emotions, which is important for social interactions. Less is known about how perceptual cues to emotions are acquired or used as individuals mature from infancy to adulthood. Here I present preliminary studies which use eye tracking and psychophysical methods to investigate the perceptual cues that infants, children and adults extract from static body and facial expressions. In a first study, we used fixation patterns and changes in pupil size to show that 7-months-old infants use cues in the upper body to discriminate fear from other body expressions (Geangu & Vuong, 2020). In a second study, we used the “bubbles” technique (Gosselin & Schyns, 2001) to compare the facial cues children (8-12 years old) and adults used to recognise emotions. The largest difference between age groups were found for fear expressions. Overall these methods can help us understand the developmental trajectory for acquiring perceptual cues to emotions.

Link to Zoom Recording of Quoc’s Talk

Suggested Readings:

Geangu, E., & Vuong, Q. C. (2020). Look up to the body: An eye-tracking investigation of 7-months-old infants' visual exploration of emotional body expressions. Infant Behavior & Development, 60, 101473.


Gosselin, F., & Schyns, P. G. (2001). Bubbles: a technique to reveal the use of information in recognition tasks. Vision Research, 41, 2261-2271.

November 10th, 2021

Amy Dawel 

Improving Face Recognition using Caricaturing


There are many circumstances in which people have problems recognising faces, including for other-race faces, as we age, or when vision is damaged. Problems recognising faces can have important real-world consequences. Here, we investigate whether caricaturing—a method predicated on theoretical ideas about face space—can be used to improve face recognition in circumstances where it is usually poor. Our data demonstrate face recognition is better for caricatured than non-caricatured images for other-race faces, images that simulate age-related macular degeneration, and in older adults. More recently, we found that six hours of training with caricatured faces of another race improved recognition of unlearned non-caricatured faces of that race (N=28 East Asians raised in Asia, viewing Caucasian faces). Together, these findings imply caricaturing may be a useful applied strategy for improving poor face recognition, including in adults who lack experience with faces of another race during the childhood sensitive period.

Link to Zoom Recording of Amy’s Talk

Suggested Readings: 

Dawel, A., Wong, T. Y., McMorrow, J., Ivanovici, C., He, X., Barnes, N., ... & McKone, E. (2019). Caricaturing as a general method to improve poor face recognition: Evidence from low-resolution images, other-race faces, and older adults. Journal of Experimental Psychology: Applied, 25(2), 256.


McKone, E., Dawel, A., Robbins, R. A., Shou, Y., Chen, N., & Crookes, K. (2021). Why the other‐race effect matters: Poor recognition of other‐race faces impacts everyday social interactions.British Journal of Psychology.


McKone, E., Wan, L., Pidcock, M., Crookes, K., Reynolds, K., Dawel, A., ... & Fiorentini, C. (2019). A critical period for faces: Other-race face recognition is improved by childhood but not adult social contact.Scientific reports, 9(1), 1-13.


October 13th, 2021 

Connor Parde 

Deep learning insights for single-unit and neural population codes in face recognition


Single-unit responses and population codes differ in the “read-out” information they provide about high-level visual representations. Diverging local and global read-outs can be difficult to reconcile with in vivo methods. To bridge this gap, we studied the relationship between single-unit and ensemble codes for identity, gender, and viewpoint, using a deep convolutional neural network (DCNN) trained for face recognition. At the unit level, we measured the number of single units needed to predict attributes (identity, gender, viewpoint) and the predictive value of individual units for each attribute. Identification remained accurate when sampling only 3% of the network’s output units, and all units had substantial identity-predicting power. In addition, cross-unit responses were minimally correlated, indicating non-redundant identity cues. Alternatively, accurate gender and viewpoint classification required large-scale pooling of units. At the ensemble level, principal component analysis of face representations showed that identity, gender, and viewpoint separated into high-dimensional subspaces, ordered by explained variance. These identity, gender, and viewpoint subspaces contributed to all individual unit responses, undercutting neural tuning analogies for both DCNNs and, by analogy, high-level vision.

Link to Zoom Recording of Connor’s Talk

Suggested Readings:


Parde, C. J., Colón, Y. I., Hill, M. Q., Castillo, C. D., Dhar, P., & O’Toole, A. J. (2021). Closing the gap between single-unit and neural population codes: Insights from deep learning in face recognition. Journal of Vision, 21(8), 15. doi:10.1167/jov.21.8.15 

September 8th, 2021 

Jim Tanaka 

Perceptual expertise: How experience changes the way we see the world


Experts perceive the world differently than novices.  For example, whereas a novice will identify a feathery creature in the bush simply as a “bird,” the expert birder will immediately recognize this same object as a “Bachman warbler”.   What are the factors that drive this “downward shift” in recognition?  How do the processes of specialized object experts compare to the general expertise processes used in everyday face recognition? In my talk, I will explore the visual strategies and neurophysiological processes of real-world experts and examine the training protocols by which a novice becomes an expert.


Link to Zoom Recording of Jims Talk


Suggested Readings: 

Campbell, A., & Tanaka, J. W. (2018). Inversion impairs expert budgerigar identity recognition: A face-like effect for a nonface object of expertise. Perception, 47(6), 647-659. PDF

Chin, M. D., Evans, K. K., Wolfe, J. M., Bowen, J., & Tanaka, J. W. (2018). Inversion effects in the expert classification of mammograms and faces. Cognitive research: principles and implications, 3(1), 31. PDF

Hagen, S., Vuong, Q. C., Scott, L. S., Curran, T., & Tanaka, J. W. (2014). The role of color in expert object recognition. Journal of vision, 14(9), 9, 1-13. PDF


bottom of page