The Fall 2021/Spring 2022 COGNITIVE SCIENCE COLLOQUIUM SERIES schedule is shown below. Details will be posted as soon as they are available. As usual, the colloquium will be held on Fridays (unless otherwise noted), from 12:00 - 1:30 p.m., in the Speech, Language, and Hearing Sciences Building, Room 205, 1131 E Second Street. Recordings of these talks are available on Panopto, which can be accessed with a UA NetID.
Since 2012, an annual feature of the colloquium series is a special talk given by the Roger N. Shepard Distinguished Visiting Speaker. Please follow the link for a list of past speakers.
If you would like to receive email announcements about these and other events, please contact Program Coordinator Kirsten Cloutier Grabo at email@example.com to be added to the colloquium listserv.
Information about previous talks during this academic year can be found at the bottom of this list. Other past talks can be found at COGNITIVE SCIENCE COLLOQUIUM ARCHIVE.
2021/22 COGNITIVE SCIENCE COLLOQUIUM SERIES
October 22, 2021
Becket Ebitz, Assistant Professor, Neuroscience, Université de Montréal
A Neurobiology of Mistakes
Abstract: We have the capacity to (1) understand and represent the value of the actions available to us and (2) to perform the computations needed to identify the most rewarding ones. However, we do not always make these best choices, even when there is no conceivable benefit to doing otherwise. Why do we make mistakes? In this talk, I will describe ongoing work that suggests that at least some mistakes emerge from fundamental constraints in neural hardware. We find that there are limits on both the stability and accuracy of neural representations that can explain at least some of our mistakes. While mistakes are certainly costly, the constraints that cause them may confer surprising advantages: they seem to increase the efficiency and flexibility of decision-making over longer time scales, even as they produce the occasional misstep in the moment.
October 29, 2021
Martina Poletti, Assistant Professor, Brain and Cognitive Science, University of Rochester
The interplay of attention and eye movements at the foveal scale
Abstract: Human vision relies on a tiny region of the retina, the foveola, to achieve high spatial resolution. Foveal vision is of paramount importance in daily activities, yet its study is challenging, as eye movements incessantly displace stimuli across this region. Building on recent advances in eye-tracking and gaze-contingent display, we have examined how attention and eye movements operate at the foveal level. We have shown that exploration of fine spatial detail unfolds following visuomotor strategies reminiscent of those occurring at larger scales. Together with highly precise control of attention, this motor activity is linked to non-homogenous processing within the foveola and selectively modulates sensitivity both in space and time. Therefore, high acuity vision is not the mere consequence of placing a stimulus at the center of gaze: it is the outcome of a synergy of motor, cognitive, and attentional processes, all finely tuned and dynamically orchestrated.
November 5, 2021
Brad Story, Professor, Speech, Language & Hearing Sciences, University of Arizona
Transformation of discrete phonetic segments into speech based on the acoustic relativity of the vocal tract
Abstract: The vocal tract is typically defined as the airway extending from the larynx to the lips. During the production of speech, a neutral and talker-specific configuration of the airway is modulated almost continuously by the movements of the tongue, jaw, lips, velum, and larynx. From an acoustic perspective, the airway can be considered to be a non-uniform conduit whose shape at a given instant of time supports a specific pattern of acoustic resonances that transmit information related to both the intended message and the identity of the talker. This presentation will summarize recent development of a model in which individual speech segments that comprise a word, phrase, or sentence are specified as relative deflections of the resonance frequencies of the neutral vocal tract configuration, and then transformed to time-dependent modulations of the airway. The output of the model is artificial speech that can be presented to listeners. Examples will demonstrate the construction of speech with the model, results from a recent perceptual experiment, and few speech illusions that may occur with constraints imposed on the vocal tract or when distortions of the speech signal occur.
November 12, 2021
Daniel Balliet, Professor, Experimental and Applied Psychology, University of Amsterdam
November 19, 2021
Anna Dornhaus, Professor, Ecology & Evolutionary Biology, University of Arizona
COLLOQUIUM SPEAKERS who have already visited 2021/22
September 10, 2021
Andrew Wedel, Professor, Linguistics, University of Arizona
The role of communication efficiency in shaping language
Abstract: Over the last century, we've gained a great deal of evidence that language structures evolve in ways that optimize communication efficiency. In the lexicon for example, Zipf (1939) famously showed that words which are more predictable tend to be shorter, and vice versa. This relationship reduces overall speaker effort while preserving communication accuracy. In the first part of this talk, I will review some of the most interesting recent findings that illustrate the apparent influence of communication efficiency on lexicons and grammars. In the second part, I will present two strands of our research in this area that are based on the fact that listeners process the speech stream incrementally, continually updating their lexical search as the phonetic signal unfolds. As a consequence, segments earlier in words contribute on average more disambiguating information to lexical access than later segments. If languages evolve to optimize communication efficiency, we expect therefore the most informative segments should be concentrated early in words where they can do the most work in lexical disambiguation - and further that this tendency should be strongest for the least predictable words where comprehension accuracy depends more on the acoustic signal. Here I'll show data from a wide range of languages that this is in fact the case: words that are on average less predictable have relatively more informative early segments, while preserving a longer tail of redundant, confirmatory segments.
Second, I'll review our recent work suggesting that the relatively low information of late segments in a word may influence the development of phonological rules which impact lexical identification. In a typologically-balanced sample of 50 languages, we find that phonological rules which neutralize lexical distinctions (e.g., word-final obstruent devoicing in German) are common at word-ends, but very rare at word-beginnings, where neutralization would more negatively impact lexical identification. Interestingly, we find this asymmetry within our dataset in languages from every family and from every region of the world, suggesting that a bias toward word-final neutralization is a strong language universal. This is what we would expect if this asymmetry stems from a basic property of human linguistic cognition.
September 17, 2021
Jennifer Savary, Associate Professor, Marketing, Eller College of Management, University of Arizona
When Payments Go Social: The Use of Person-to-Person Payment Methods Attenuates the Endowment Effect
Abstract: Decades of research have documented a robust pattern known as “the endowment effect,” such that sellers often demand more to relinquish a good than buyers are willing to pay to acquire the same good. However almost all research to date on this phenomenon has used traditional payment methods, such as cash. In this project we examine how Person-to-Person (P2P) payment methods (e.g. Venmo, Zelle, Paypal) affect consumers’ pricing decisions in the context of the endowment effect. Building on theories related to mental associations and social norms, we predict that when consumers use P2P payment methods, they are more likely to converge on a mutually acceptable price, and the endowment effect will attenuate. This occurs because over time, people develop a set of beliefs and norms about transacting with others in a social context. P2P payment methods influence pricing decisions by inadvertently evoking these social norms and subtly prompting consumers to take into account the perspective of their transaction partner. As a result, people make somewhat less selfish, more cooperative pricing offers: sellers accept somewhat lower prices, and buyers pay somewhat more, attenuating the endowment effect. Importantly, this occurs even though people explicitly report that the transaction is with a stranger they do not know. Seven studies using consequential and hypothetical choices test this proposal and conceptual framework. Finally, a market simulation indicates that P2P can increase successful transactions in a marketplace by more than 15%.
September 24, 2021
Katalin Gothard, Professor, Physiology, University of Arizona
How the amygdala may turn expectations into emotional experience.
Abstract: The amygdala plays a central role in emotion and social behavior, yet its role in processing social and affective touch has not been established. Tactile stimuli, processed initially by the somatosensory cortex, acquire affective salience downstream from early processing stages possibly in the amygdala. We monitored simultaneously neural activity in the somatosensory cortex and the amygdala of monkeys that received alternating blocks of either innocuous, gentle air puffs or grooming-like touch from a trusted trainer. We expected neurons in the somatosensory cortex to encode the physical features of puff and touch stimuli whereas neurons in the amygdala were expected to differentiate between the neutral, non-social puff and the pleasant, social touch. The pleasantness of touch was inferred from the autonomic state of the recipients. During grooming blocks, monkeys appeared less vigilant, closed their eyes, had lower heart rates, and increased vagal tone. In contrast, during periods of puff delivery, high levels of vigilance and sympathetic arousal were evident. Surprisingly, during grooming, neurons in the amygdala stopped responding to tactile stimuli, even if the stimuli were delivered to the same areas of the skin that showed reliable responses to puff. This suggests the presence of a gating mechanism in the amygdala. Instead of responding to each touch stimulus, a set of amygdala neurons signaled with sustained changes in baseline firing rate throughout the touch blocks. These finding suggest that while receiving affective touch, the amygdala may be decoupled from monitoring the external environment, while tonically signaling to the rest of the brain the social-behavioral context and affective state of the recipient.
October 1, 2021
Cameron Buckner, Associate Professor, Philosophy, University of Houston
Imagination and the Prospects for Empiricist AI
Abstract: In current debates over deep-neural-network-based AI, deep learning researchers have adopted the goals of philosophical empiricism and associationism, and deep learning’s critics have redeployed arguments from philosophical rationalism and nativism. One of the most influential arrows in the rationalist quiver is summarized by Fodor’s claim that the ability to create new compositional representations is required for cognition. In a centuries-displaced debate, Fodor applauds Hume for acknowledging this burden, but criticizes Hume for appealing to the imagination to discharge it. Fodor claims that an associationist appealing to the imagination constitutes “cheating”, and notes that Hume never explains how the empiricist imagination actually works. More recently, deep learning researchers have claimed that generative deep neural network models (such as Generative Adversarial Networks, Variational Autoencoders, and Transformers) can perform one or more of the roles ascribed to the imagination by cognitive psychology and neuroscience. In this talk, I canvass these models and their achievements to arbitrate this dispute between Humean empiricism and Fodorian rationalism, in the process extracting more general lessons about empiricist cognitive architecture and the prospects for deep-learning-based AI.
October 15, 2021
Vinodkumar Prabhakaran, Senior Research Scientist, Google, LLC., San Francisco
Research Affiliate, Stanford University
NLP and Society: Undesirable Societal Biases as Barriers to Those in the Margins
Abstract: As natural language processing (NLP) techniques are increasingly being used in various day-to-day applications, there is growing awareness that the decisions we as researchers and developers make about our data, methods, and algorithms have immense impact in shaping our social lives. In this talk, I will outline a growing body of research on ethical implications of NLP technologies, especially around fairness failures along various axes. I will discuss ways in which machine learned NLP models may reflect, propagate, and sometimes amplify social stereotypes about people, potentially harming already marginalized groups. I will cover research from our team at Google, as well as the larger research community on ways to detect and address these issues, and discuss the open challenges in this space.