|
The Cognitive Lunch talks will be on Wednesdays from 12:10 pm - 1:25 pm in the Psychology conference room (PY 128) located behind the main office.
Abstract 9/7: Seth Frey
Dynamical behavioral game theory: evidence for a limit cycle regime in rock-paper-scissors experiments.
Even though some mathematical game theorists have explored higher dimensional solution concepts, the Nash equilibrium and other fixed points continue to dominate thinking in experimental economics. This persists even in the face of hints that they are failing. I will present a theoretical argument and behavioral evidence that subjects playing a repeating generalization of rock paper scissors are at or near a cyclic strategy set. This work demonstrates the importance of dynamical perspectives on economic games---not just for studying learning, but for understanding equilibrium.9/14: Grit Herzmann
Expertise and Recognition Memory
Expertise is well known to enhance memory, but the underlying neural processes of experts’
memory advantage are still poorly understood. I will present data from two research projects that use behavioral measures together with event-related potentials
(ERPs) to understand how
perceptual expertise facilitates recognition memory. The first project investigates the exceptional memory of car experts. The results show that higher levels of expertise lead to less brain activation during memory encoding yet also to the retrieval of more detailed information during recognition. The second project, which combines my interests in faces and expertise, investigates the other-race effect in face recognition. This project shows that the neural processes of memory encoding and retrieval that underlie superior memory performance for own-race faces are very similar to those found in expertise with other visual categories like cars.9/21: Brendan Johns
Generating structure from experience: Memory as the basis of language
Theories of language have generally assumed that abstraction of the linguistic input is necessary in order to create higher-level representations of the workings of a language (i.e. a grammar).
However, the importance of individual experiences with language has recently been emphasized by both usage-based theories (Tomasello,
2003) and grounded and situated theories (e.g. Zwaan & Madden, 2005).
Based upon this, a formal exemplar model of language is described, which stores instances of sentences across a natural language corpus, using recent advances from models of semantic memory. This memory store is used to generate expectations about the future structure of sentences, based upon the importance of prediction in language processing (Altmann & Mirkovic, 2009). This model can successfully capture a variety of behavioral effects that no other general learning-based model can simulate, such as reduced relative clause processing (Reali & Christiansen, 2007), the role of contextual constraint (Rayner & Well, 1996), and event knowledge activation (Ferretti, et al., 2007), among others. Additionally, how perceptual knowledge can be integrated into this framework, in order to ground the processing of language into perception, is described. This work provides evidence that much of language processing may be bottom-up in nature, based upon the storage and retrieval of individual experiences with language.9/28: Melissa Gresalfi
Tacit knowing to explicit explanation: Mining student designs for evidence of systems thinking
Although there is increasing awareness of the importance of engaging students in the act of producing artifacts and representations of ideas, rather than simply consuming formal, externally-produced designs, such production brings its own complexities. One key challenge of projects that engage students centrally in the practices of design involves ensuring that the core content underlying students¹ designs move beyond tacit understandings to explicit and tangible understandings. In this talk, Dr.
Gresalfi will explore the question of what can be said about what students know when they are engaged in design tasks whose successful execution requires understanding of fundamental concepts of systems thinking.
Drawing on data of students designing videogames, this talk will explore what nascent understanding of systems looks like and how these understandings relate to more formal explanations of systems.10/5: Gregory Cox
A Dynamic Model for Recognition Memory
Traditional models of recognition memory assume that recognition decisions are based on a static value of familiarity that reflects the match between a test item and the contents of memory: An item is recognized as old when it results in a familiarity value higher than some criterion. This approach becomes untenable, however, when the items to be recognized vary widely in their absolute familiarity as in the case of, say, a random dot pattern (low overall familiarity) and a high-frequency word (high overall familiarity). In addition, the traditional model offers no principled way of predicting reaction times. I will present the results of an experiment (carried out with Stephen Denton) that show that, even when stimuli vary widely along many dimensions, participants are able to make sensible recognition decisions that cannot be explained by comparison to a fixed criterion.
These results are used to motivate a general model of recognition memory that Rich Shiffrin and I have been developing. In this model, decisions are based not on a fixed familiarity value, but on a dynamic activation value that evolves over time as information is extracted from the test item and compared to memory. Changes in this dynamic activation are fed into two racing accumulators, one leading to an "old" decision, the other to a "new" decision, where the "winner"
determines the predicted response and decision time. These joint predictions of accuracy and decision time are in accord with benchmark findings in the recognition memory literature, as well as the results of our experiment.10/12: Ehtibar Dzhafarov
Causality, Probability, and Selectivity in Psychology and Beyond
Given a set of several inputs into a system (e.g., independent variables characterizing stimuli) and a set of several stochastically non-independent outputs (e.g., random variables describing different aspects of responses), how can one determine, for each of the outputs, which of the inputs it is influenced by? In psychology, the problem has applications ranging from modeling pair-wise comparisons to reconstructing mental processing architectures to conjoint testing. A necessary and sufficient condition for a given pattern of selective influences is provided by the Joint Distribution Criterion, according to which the problem of "what influences what" is equivalent to that of the existence of a joint distribution for a certain set of random variables. While new in the behavioral context, the Joint Distribution Criterion has been previously invoked in quantum physics, in dealing with generalizations of Bell inequalities for the quantum entanglement problem. The parallels between this problem and that of selective influences in behavioral sciences is established by observing that non-commuting measurements in quantum physics are mutually exclusive and can therefore be treated as different levels of one and the same factor.10/19: Dr. Shahzeen Z. Attari, Assistant Professor, School of Public and Environmental Affairs, Indiana University
Human behavior and energy consumption – Understanding decisions about energy
Understanding the relationship between human behavior and energy use is vital to decrease per capita energy consumption. In this presentation, I summarize my recent study where participants reported their perceptions of energy consumption for a variety of household activities. When asked for the single most effective strategy they could implement to conserve energy, most participants mentioned curtailment (e.g., turning off lights) rather than efficiency improvements (e.g., installing more efficient light bulbs), in contrast to experts’ recommendations. Participants had small overestimates for low-energy activities and large underestimates for high-energy activities. Follow-up studies show that participants find implementing recommended changes to decrease energy consumption relatively easy and that they want to incorporate easier non-effective behaviors for themselves and expect others to implement harder more-effective behaviors. Future research that stems from these studies include (1) a further look at what motivates and demotivates action in social dilemmas, where private interests are at odds with collective interests, (2) a real-time energy feedback project aimed to correct misperceptions of energy consumption and facilitate behavior change, and (3) a game that allows participants to learn about energy-saving behaviors and experience climate change impacts viscerally.10/26: Carolina Tamara
Linking Egocentric and Geocentric Frames of Reference in Navigating Rats
Animals collecting renewable resources are faced with complex spatial problems.
In a familiar terrain, most species can return to those resources by using either egocentric or geocentric systems of reference. In the first case, they continuously monitor their path by tracking the direction and distance they have covered (i.e. path integration). In the second case, they rely on their memory of landmarks and environmental boundaries.
I will present a series of studies examining how foraging rats rely on a combination of learned routes and recognition of visual landmarks. I will also discuss how these systems can support navigation without a map-like representation of the space and the similarities in the navigational strategies of birds and mammals.11/2: Melody Dye
No Representation Without Taxation
How do the ways in which we learn influence our representations of what we learn? In this talk, I will present the results of a series of simulations and experiments with Michael Ramscar and Dan Yarlett that show how the process of learning to conceptualize and categorize perceptual input indelibly shapes how that input gets represented in mind. In representation, there seems to be a give and take between veridicality and completeness, on the one hand, and discrimination and accurate categorization, on the other. Learning to sort objects into categories based on their highly-diagnostic features makes people less likely to notice or remember the same objects' less-diagnostic features. Gains in response-discrimination between categories thus come at a cost to within category discrimination. These findings suggest that the mechanisms of human learning obey a simple principle:
there can be no representation without taxation.11/9: Richard Prather
Numerical Development: From Percept to Concept
The current presentation includes computational and behavioral evidence that demonstrates how the neural coding of numerical magnitude influences the development of number cognition. Behavioral phenomena are often attributed to some type of cognitive-representation, such as “logarithmic number representations”
or the mental number line. I present evidence from a series of computational models that suggest that the neural coding of number, and how it changes with development may provide a parsimonious mechanism for these behavioral phenomena. I will address both the development of number-line estimation and operational momentum.
Additionally, I present behavioral and computational work that addresses the consequences of several hypotheses relevant to numerical development, such as General Magnitude (Walsh, 2003), cross-modal interactions and the effects of label use.11/16: Matthew Hurley
A Funny Way out of the Frame Problem--Seriously: Where's the sense in humor?
The frame problem is an epistemic predicament for any finite cognitive agent that needs to make behavioral decisions in a complex world. When considering the world, how does an agent choose which thoughts to have, and which of the innumerable possible thoughts to exclude? What is relevant and what is irrelevant? As humans we face this problem, and A.I. engineers also face it when designing their agents. Many facets of current cognitive science and psychology seem to be converging on a somewhat reasonable heuristic solution to the frame problem. I'll describe the way my colleagues and I see this solution at a cognitive level in terms of working memory spaces that are furnished on the fly by relevant semantic contents via spreading activation inference. Then I'll note that such a solution, while perhaps indispensable for a useful cognition, necessarily leaves a hole behind--a smaller epistemic failure. Part, though not all, of this hole can be plugged. Lastly, I'll describe a hypothesis in which humor plays the role of this incremental epistemic plug.11/30: Chi-hsin (Esther) Chen
Prior Language Experience Affects Statistical Learning of Object Names and Categories: Evidence from Cross-Linguistic Studies of Adult Learners
Previous studies have shown a close link between language learning and object categorization. Both adults and children are able to use object labels and abstract linguistic structures (e.g., grammatical gender in gender-marking languages) as a cue in object category learning and generalization. In this talk, I will present a series of experiments that investigate the effects of the morphological structures of a language on its speakers’ sensitivity to different types of statistical regularities in novel stimuli.
Three experiments were conducted to examine whether native speakers of English and Mandarin performed differently when exposed to training data that either reflected, or were inconsistent with, the linguistic features in their native language. With the same statistical regularities in the dataset, participants performed better when the training stimuli reflected the linguistic structure of their native language. In addition, the results suggest that word learning and object categorization serve to bootstrap each other. That is, learners are able to utilize information from newly formed categories to acquire additional word-referent mappings in a bi-directional and mutually supporting relationship.12/7: Richard Veale
Vision, Eye Movements, and Habituation in the Human Newborn (and Beyond)
Humans, as primates, are born with their eyes open and actively moving. Their neural and mechanical systems are sufficiently mature to not only spontaneously and selectively direct the eyes among complex visual environments, but also to adapt the allocation of this looking based on visual experiences. This adaptive learning behaviour is visual habituation. The subject of this talk is the current understanding of the rules that govern visual habituation and looking behaviour, and how these are implemented by the neuromechanical system comprising the human neonate. This is accomplished with reference to my research building neuro-robotic models of the visuo- and oculomotor systems of human infants. These robot infant surrogates participate in behavioral experiments to match behavioral measures from real infant and primate experiments, with the advantage that we have at all times full enumeration of the state of the subject's neural system, body, and environment. This exhaustive knowledge enables analysis of dynamics and interactions within and between the brain, body, and environment to understand how the system works mechanistically.
This talk will address the current state of the research, including enumerating the visual habituation/looking behaviour of newborn humans, and construction of the neuro-robotic models of visuo/oculomotor behaviours based on neurophysiological and anatomic literature. Recently obtained preliminary results will also be presented regarding success matching looking behaviour. The talk will conclude with discussion on modelling infant development after birth, and how this research can inform understanding of more general and useful behaviours such as multimodal audio-visual habituation in dynamic interactions with humans (which we have already had some success modelling), and eventually language learning.
Previous Cognitive Lunch
|
