CHAZ FIRESTONE (Psychology, Johns Hopkins University)
TAKING A MACHINE'S PERSPECTIVE
Friday, January 25, 2pm, BSB 163 (this is a joint talk with the Centre for Vision Research)
How similar is the human mind to the sophisticated machine-learning systems that mirror its performance? Convolutional Neural Networks (CNNs) have taken our field by storm, achieving human-level benchmarks in recognizing novel images and objects. These advances support transformative technologies such as autonomous vehicles and machine diagnosis, but beyond this they also serve as candidate models for the human mind itself -- not only in their output but perhaps even in their underlying mechanisms and principles. However, unlike humans, CNNs can be "fooled" by adversarial examples -- carefully crafted images that appear as nonsense patterns to humans but are recognized as familiar objects by machines, or that appear as one object to humans and a different object to machines. This seemingly extreme divergence between human and machine classification challenges the promise of these new advances, both as applied image-recognition systems and also as models of the human mind. Surprisingly, however, little work has empirically investigated human classification of such stimuli: Does human and machine performance fundamentally diverge? Or could humans engage in some “machine theory of mind” and predict the CNN’s preferred labels? Here, I’ll show how human and machine classification of adversarial stimuli are surprisingly related: I will present data showing that, across many prominent and diverse adversarial imagesets, human subjects can reliably identify the machine's preferred label over relevant foils, even for images described in the literature as "totally unrecognizable to human eyes". I suggest that human intuition may be a more reliable guide to machine (mis)classification than has typically been imagined, and explore the consequences of this result for minds and machines alike.
MAGGIE TOPLAK (Psychology, York University)
ASSESSING THE DEVELOPMENT OF COGNITIVE ABILITIES AND COGNITIVE BIASES
Wednesday, February 6, 12:30pm, Ross S421
Many cognitive abilities show a steady increase throughout childhood and adolescence. However, previous research has found that performance on some cognitive biases (such as heuristics and biases tasks) show improvement with age, but others do not. The developmental course of cognitive biases remains largely unknown. It is particularly challenging to identify suitable stimuli for the assessment of cognitive biases in developmental samples, given that these paradigms were originally developed using adult samples. In addition to the developmental suitability of these items, an additional challenge of using developmental samples is the rapid, parallel development of general cognitive abilities such as intellectual abilities and executive functions. In order to advance our understanding of the development of cognitive biases, we have examined their association with different indicators of cognitive sophistication. In our program of research, we have examined performance on cognitive abilities and cognitive biases cross-sectionally across different periods of development. We have also examined whether cognitive abilities and dispositional tendencies that support rational thinking are correlated with resistance to cognitive biases. Our work has demonstrated that children and youth who display resistance to cognitive biases tend to display higher cognitive abilities and tendencies toward actively open-minded thinking. Most recently, we have conducted a longitudinal follow-up of our original developmental sample to examine developmental trajectories of these measures. I will also report on the findings from a cohort-sequential longitudinal design of typically developing children and youth. Our sample spans the range from 8 to 20 years of age based on testing at three time points, each separated by three years. We estimated latent growth curve models to examine the developmental trajectories of resistance to cognitive biases, based on a composite measure including baserate sensitivity, ratio bias, belief bias syllogisms, resistance to framing and temporal discounting. We also estimated these models for intellectual abilities (verbal and nonverbal), executive functions (interference control and mental flexibility), and actively open-minded thinking (AOT). Together, our results provide further evidence for the development of resistance to cognitive biases and convergence with other indicators of cognitive sophistication. These results also highlight the role of individual differences for understanding how children and youth improve or fail to show improvement on resisting cognitive biases.
MUHAMMAD ALI KHALIDI (Philosophy, York University)
NEURAL CORRELATES WITHOUT REDUCTION: THE CASE OF THE CRITICAL PERIOD
Wednesday, February 27, 12:30pm, Ross S421
Researchers in the cognitive sciences often seek neural correlates of psychological constructs. In this talk, I argue that even when these correlates are discovered, they do not always lead to reductive outcomes. To this end, I examine the psychological construct of a critical period and briefly describe research identifying its neural correlates. Although the critical period is correlated with certain neural mechanisms, this does not imply that there is a reductionist relationship between this psychological construct and its neural correlates. Instead, this case study suggests that there may be many-to-many psychological-neural mappings, not just one-to-one or even one-to-many relations between psychological kinds and types of neural mechanisms.
STEVEN PIANTADOSI (Psychology, University of California Berkeley)
ALGORITHMIC INFERENCE AS THE BASIS OF HUMAN LEARNING
Wednesday, March 27, 12:30pm, Ross S421
I'll present an overview of my research that is aimed at understanding how human learners solve complex, structured learning problems. Recent theories of human learning have hypothesized that people can infer the algorithm or computation giving rise to the data they can observe. This approach shows promise in explaining human behavior across a variety of domains, including language learning, number acquisition, and conceptual development generally. It also allows the field to address even more basic questions about what types of knowledge might be "built in" for humans, and how children develop the rich systems of knowledge found in adults. I'll describe a series of studies on mathematical learning and cognitive development in children, US adults, and indigenous Amazonians, and describe families of computational models that we can use to capture the remarkable statistical inferences carried out by human learners.
Past Speaker Series
Sep 12—Laura Niemi (Munk School, University of Toronto), Tracing the Process of Moral Judgment in Language and Cognition
When things go wrong, people ask questions like: “Who made it happen?” “Who was responsible?” and “Who is to blame?” In other words, they engage in a process of moral judgment that involves causal cognition. To what extent is this process permeated by people’s diverse values and ideological commitments; and, to what extent is it influenced by the language used to describe the event? This talk will cover findings from several studies combining individual differences measures with vignette-based experiments and psycholinguistics tasks. Collectively, the research demonstrates that values systematically map onto different patterns of causal attribution and language use. Studying morality through the lens of language brings precision to our understanding of the psychological underpinnings of diverse values, and also indicates that our understanding of language is incomplete without consideration of moral psychology.
Oct 3—Dale Stevens (Psychology, York University), Are Object Concepts Hardwired in the Brain?
Discrete parts of the human brain respond preferentially to very particular categories of objects. Moreover, the general organization of these “category-specialized” brain regions is remarkably similar across individuals. This is one of the most robust and oft replicated findings in the field of cognitive neuroscience, and it has sparked much controversy and debate regarding the fundamental nature of object concepts in the brain. Is this category-related organization innate/hardwired in the brain, or driven solely by external perceptual characteristics of objects, or something else? In this talk, I present recent evidence from my neuroimaging research demonstrating that while stable anatomical connectivity constrains the spatial topography of this category-related organization, malleable experience-driven "functional connectivity" among brain regions gives rise to category-specilazation.
Oct 24—Adam Pautz (Philosophy, Brown University), How Does Experience Represent the World?
Like many others, I think we should accept a representational theory of sensory consciousness: being conscious of the world is a matter of representing the world. Thus the hard problem of consciousness becomes the hard problem of representation. For example, how do electrical events in soggy grey matter enable us to represent bright orange pumpkins? The most popular answer (Armstrong, Dretske, Tye, Byrne and Hilbert, Hill) is that the sensible qualities (colors, smells, tastes, etc.) are "in the world" and the brain represents them by undergoing states that have the biological function of indicating them. In this talk, my primary aim is to develop some empirical arguments against this view. At the end, I will briefly motivate a radically different approach: a kind of internalist, non-reductive form representationalism.
NB: Pautz will be giving a companion talk the next day at the University of Toronto in which he will develop some quite different, more a priori arguments for a non-reductive, internalist theory of consciousness. The arguments will be founded on a series of novel thought-experiments.
Nov 7—Hayley Clatterbuck (Philosophy, University of Rochester), How Does Language Create New Concepts?
Compared to other species, humans seem to have an exceptional capacity for representing and learning abstract concepts. According to the “language-first hypothesis”, language – and not some antecedent change in our representational abilities – explains how we first gained these abilities and how individuals today learn new concepts. To test this hypothesis, I consider whether and how language can play this role, drawing on Carey’s bootstrapping account and several techniques from machine learning. Finally, I investigate whether associative mechanisms found in other species suffice for the kind of word learning that creates new abstract concepts.
Nov 03, 2017, at 3.30 pm
Geoffrey MacDonald (Psychology, University of Toronto)
Love is the Drug: Social Reward and Interpersonal Behaviour Regulation
Abstract: Although a general principle is that animals regulate behaviour based on avoiding punishments and approaching rewards, relationship science has largely focused on safety rather than reward motives. In this talk, I argue for the importance of reward in the regulation of interpersonal behaviour. The studies I will discuss show that people regret missing opportunities for social reward and pursue relationships that promise reward. My data suggest that social reward is mediated by the release of endogenous opioids reflecting its addictive qualities. Finally, I explore boundary conditions to the pursuit of reward such that individuals high in the fear of being single or attachment avoidance are less motivated by social rewards.
Time and location: 3.30-5.30 pm (Friday), Ross S 421
Oct 14 (Fri) with Philosophy: Sharon Street (NYU, Philosophy)
Meditation, Metaethics, and the View from Everywhere
Oct 19: Daphna Buchsbaum (UoT, Developmental Psychology)
How do you know that? Integrating Causal Knowledge and Learning from Others
We live in a causally complex world, where we must learn not only to predict the consequences of events (“the wind blowing could make that branch fall on me”), but also to act causally on the world ourselves (“pressing the remote control button turns on the TV”). How do children learn causal relationships, especially when the world presents them with sparse, ambiguous data or with multiple, conflicting sources of evidence? Social learning may be especially beneficial —with little expertise and few life experiences, children can quickly acquire large amounts of new information from other people without spending the time and effort to learn through trial-and-error. However, not all information from others is equally dependable. People can be ignorant, make mistakes, or give conflicting information. I will first present work suggesting that children are able to rationally combine multiple sources of information about which actions are causally necessary when deciding what to imitate, interpreting the same statistical evidence differently when it comes from a knowledgeable teacher versus a naïve demonstrator. I will next present research looking at how children and adults combine direct observation of probabilistic data with causal predictions provided by a social informant, and how this influences their future trust in that informant. Finally, I will present research looking at how people reconcile differences in opinion amongst multiple demonstrators, including how they balance the opinion of a majority against the quality of informants’ information. Throughout this work, I use computational probabilistic models to evaluate what learners with differing social assumptions should rationally infer from the social and statistical evidence they receive.
Oct 26: Chris Westbury, Cognitive Psychology, Linguistics)
Beyond ‘takete’ and ‘maluma’: Using big data to understand sound symbolism
Sound symbolism is the phenomenon of extracting semantics from formal (orthographic and/or phonological) elements of a string. Köhler (1929/1947) famously showed that people were much more likely to associate the nonword ‘takete’ with a spiky shape and the nonword ‘maluma’ with a round shape than the other way around. Sapir (1929) showed that people were more likely to associated the string ‘mal’ than the string ‘mil' with large things. These findings have been much replicated: indeed, a large proportion of the sound symbolism literature (40% in a review of 99 studies) consists of follow-up studies to Köhler and Sapir. I will point out several limitations in the sound symbolism literature and present results from three recent studies that try to overcome these limitations by using ‘big data’ (experiments that use thousands of randomly-generated stimuli). The first two studies address an unusual question that turns out to have a surprisingly clear and simple answer: Why do people consistently find some nonword strings humourous? The third study characterizes sound symbolic effects in nearly two dozen semantic categories, including several for which no sound symbolism effects have ever been suggested. I will end by discussing several plausible reasons why sound symbolic effects exist, and what their existence suggests about human cognitive processing.
Paul Katsafanas (Boston, Philosophy): "Fanaticism and Sacred Values"
Luke Roelofs (Philosophy, Australian National University)
'Octopuses, split-brains, and the universe: how unified does consciousness have to be?'
Short abstract: Normal human consciousness is in many ways remarkably well-integrated; plausibly this is part of what leads us to think of each person as a single conscious subject. By contrast, the conscious goings-on in the universe as a whole are not similarly well-integrated; plausibly this is part of what leads us to think of them as not belonging to a single conscious subject. But what should we think about systems that seem to fall somewhere in between, displaying too much integration to be called simply many and too little to be called simply one? With an eye to two particular examples of such cases (cephalopods and callosotomy patients) I review some rival ways of thinking about this question, and consider how far we can retain the simplicity of the common-sense outlook.
Jacob Beck (Philosophy, York University)
‘Is sensory experience analog?’
Abstract: Back in the 1980s several philosophers argued, on broadly introspective and a priori grounds, that sensory experience is analog. In the ensuing years, these arguments have been forcefully criticized, leaving the thesis that sensory experience is analog in doubt. My talk will have two aims: to diagnose a common flaw in these past arguments that traces to their armchair methodology; and to begin to develop a new, and more empirically informed, argument for the same conclusion.
Lunch and refreshments will be provided at the talks.
Wednesday, November 18
Jennifer Steele (Psychology, York)
“How and When Do Children's Implicit Racial Biases Develop?”
Wednesday, December 02
Tim Bayne (Philosophy, University of Manchester and Western University)
"Can we Build a Consciousness Meter?
September 10, 2014
Otavio Bueno (Philosophy, University of Miami)
“What Does a Mathematical Proof Really Prove?” *
September 24, 2014
Joni Sasaki (Psychology, York University)
“The Cultural and Biological Shaping of Religion's Effects”
October 8, 2014
Tina Malti (Psychology, University of Toronto, Mississauga)
“Mind, Emotions, and Morality”
October 22, 2014
Serife Tekin (Philosophy, Daemen College)
“Against Grief Erosion: Incompatible Research and Clinical Interests in Psychiatric Taxonomy”
November 19, 2014
Laurence Harris (Psychology and CVR, York University)
“The Vestibular System and the Sense of Self”
Wednesday, January 28
Frank Russo (Psychology, Ryerson)
"Oscillatory Brain Dynamics Underlying the Perception of Pitch, Rhythm,
and Emotion in Music and Speech"
Friday, February 6*
Ernie Lepore (Philosophy and Cognitive Science, Rutgers)
"On the Perspective-Taking and Open-Endedness of Slurring"
Wednesday, February 25
Robert Foley (Philosophy, Western)
"Flexible Interaction as a Criterion for Consciousness"
The speaker series is held on Weds at 3:30 pm in Ross S 421, unless otherwise indicated.
* = Joint session with Philosophy Department Colloquium
September 18, 2013
Keith Schneider (Biology and Centre for Vision Research, York University)
"Visual Attention Affects our Decisions but not Perceptions"
October 16, 2013
Wayne Wu (Philosophy and Center for Neural Basis of Cognition, Carnegie Mellon University)
"What is Attention?"
January 31, 2014*
Robert McCauley (Philosophy, Emory University)
"The Cognitive Foundations of Science and Religion"
February 26, 2014
Steven Sloman (Psychology, Brown University)
"Explanation Fiends and Foes: Different Modes of Causal Reasoning"
March 12, 2014
Tina Malti (Psychology, University of Toronto - Mississauga)
"Mind, Emotions, and Morality" -- (Postponed due to weather)
* Joint with Philosophy Department Colloquium, scheduled for Friday instead of Wednesday.
September 19, 2012
John Heil (Philosophy, Washington University St Louis)
October 17, 2012
Adam Cohen (Psychology, Western)
“Theory of mind as a cognitive reflex”
November 7, 2012
Louise Barrett (Anthropology, Lethbridge)
“A little less representation, a little more action, please”
January 30, 2013
Rebecca Saxe (Neuroscience, MIT)
"The Happiness of Fish: Neural Mechanisms for Understanding Minds Unlike Your Own"
March 6, 2013
Hakob Barseghyan (Institute for History and Philosophy of Science, University of Toronto)
"A Descriptive Theory of Scientific Change"
April 5, 2013
Tyler Burge (Philosophy, UCLA)
"Perception: Origins of Mind"
Interdisciplinary Workshop - Animal Pain and Consciousness
January 11, 2013
Colin Allen (Department of Philosophy, University of Indiana)
Kristin Andrews (Department of Philosophy, York University)
Verena Gottschling (Department of Philosophy, York University)
Suzanne McDonald (Department of Psychology, York University)
Anne Russon (Department of Psychology, York University, Glendon)
Adam Shriver (The Rotman Institute of Philosophy, University of Western Ontario)
- Fall 2011-2012 (PDF)
- Winter 2011-2012 (PDF)
- Fall 2010-2011 (PDF)
- Winter 2009-2010 (PDF)
- Fall 2009-2010 (PDF)
The Cognitive Science program also sponsors workshops on various topics in cognitive science, bringing together philosophers, psychologists, neuroscientists and others to discuss their latest research: