back to seminar listings

Institute for Cognitive and Brain Sciences (ICBS) UC, Berkeley

Fall 2004 Colloquium Series:

Colloquia are at 11 am in Tolman 5101 (unless noted by **)

Sept 17 Dan Dennett,
Tufts University, Center for Cognitive Studies,
"The Personal Level and the Decomposition of Qualia."
Oct 1 Kaiping Peng,
UC Berkeley, Department of Psychology
"Culture and Cognition"
Oct 15 Michael Ramscar
Stanford University, Department of Psychology
"How children avoid the logical problem of language acquisition."
Oct 29 Shlomo Bentin,
Hebrew University,
Department of Psychology and Center for Neural Computation
"Semantic impact on forming new associations: From experimental psychology to neural networks, ERPs and back."
11/05 Cameron Carter,
UC, Davis, Department of Psychiatry and Psychology
"How is cognitive control controlled?"
11/19 Krishna Shenoy,
Stanford University,
Department of Electrical Engineering & Neurosciences Program
"Neural Basis of Reach Preparation and Motor Prosthetics."
12/03 David Wessel,
UC, Berkeley
Department of Music and Center for New Music and Audio Technologies
"Musical Perceptions and Musical Actions"
12/10 Tom Griffiths
Stanford University
Department of Psychology
"Theory-Based Causal Induction"

Abstracts

Dennett

"The Personal Level and the Decomposition of Qualia."

The problems of "imagination management" are severe when one attacks the problem of consciousness. On the one hand, we can imagine things at the personal level, at which we find selves, qualia, pains, intentions, beliefs, and the other familiar furniture of folk psychology. On the other hand, we can imagine things at any one of several sub-personal levels, where all these familiar items disappear, to be replaced, somehow, by patterns of neuronal activity. Several stumbling blocks from philosophy have mesmerized cognitive scientists into mis-imagining these issues. Something like visualization therapy is called for, and will be provided in the form of several exercises for the imagination.

Ramscar

"How children avoid the logical problem of language acquisition."

In learning a language, children need to generalize beyond the available data. Logically, the way children seem to make these generalizations appears to seriously undermine empiricist accounts of language acquisition. If children make generalizations that both encompass and go beyond the language they are acquiring ("superset hypotheses"), any positive evidence they encounter will support their error, and thus they will require negative feedback if achieving the correct pattern of behavior is dependent only on the input received. Since children make superset hypotheses, and neither receive nor respond to negative feedback, the case for language specific constraints has seemed overwhelming. In this talk, I'll question this seemingly implacable logic, and outline a learning model that suggests that children may in fact shake off errors and show mastery of grammatical forms simply by repeating their erroneous productions. I will then describe experimental evidence in support of this model. In these experiments, children manifest learning of correct linguistic forms in a situation where, as a result of philosophical and linguistic analyses, it has been argued that it is logically impossible for them to do so.

Bentin

"Semantic impact on forming new associations: From experimental psychology to neural networks, ERPs and back."

Co-occurrence is the most widely accepted factor on forming associations among events. The rules governing this process have been extensively investigated since Socrates and Aristotle thinking, the Britsh Empiricist School and the behaviorist psychologists. In the vast majority of these studies, however, associative learning was intentional and rather deployed of context. In real life, however, associations are usually form incidentally, and co-occurrence is never arbitrary. In the present study, my students and I used traditional experimental psychology and ERP designs to explore the role of semantic relationship and sentential context in forming new episodic associations between words. We used computational techniques to model the interaction between the semantic and episodic networks in forming associations, and successfully replicated the empirical results in computer simulations using that model. Further, the model has predicted new characteristics of the associative dynamics that have been validated back in the psychological laboratory. Within the imposed time limits of one talk, I will tell you this story.

Carter

"How is cognitive control controlled?"

Functional neuroimaging and neuropsychological studies have implicated a distributed neural network, including the dorsolateral prefrontal cortex (DLPFC), anterior cingulate cortex (ACC) and parietal lobes, in the implementation and coordination of cognitive control. However the functional specialization of components of this network remains controversial. Control-conflict loop theory hypothesizes that the region of the ACC, on the medial frontal wall, monitors for the occurrence of conflicts during information processing while the DLPFC implements control through the representation and maintenance of task context. Interactions between these subregions of the frontal cortex are hypothesized to account for the dynamic adjustments in cognitive control that characterize normal performance. A series of studies using event-related fMRI and high density ERP recordings will be reviewed that test predictions of this model, together with a discussion of their more general implications for understanding the human cognitive architecture and mechanisms underlying cognitive disorders.

Shenoy

Department of Electrical Engineering & Neurosciences Program

"Neural basis of reach preparation and motor prosthetics" Voluntary actions are initiated more slowly than might be expected. This additional time is presumed to be used to accurately prepare the movement. As an example, an animal may have only one opportunity to strike at its prey. As movement speed and accuracy are of foremost importance, there is value in preparation even if reaction time is increased slightly. Despite the importance of such motor preparation, we do not yet understand its underlying neural mechanism. I will describe our recent electrophysiological recording and microstimulation experiments in monkeys trained to perform a delayed reaching task, which are revealing a neural correlate of movement preparation that is capable of predicting reaction time. I will also describe our recent experiments aimed at substantially increasing the performance of cortically-controlled neural prosthetic systems.

Wessel

Musical Perceptions and Musical Actions

The perception-action link in audition and musical practice is explored from a multi-sensory perspective. An enactive approach to the exploration and mastery of musical material is developed. The argument weaves together experimental evidence from both behavioral and neuroscience studies, notions of conscious control and automaticity in musical performance, and the concept of active listening. Finally the design of computer-based musical instruments that provide for intimate control and the expressive acting-out of music is considered. Examples of such composed instruments are demonstrated.

Griffiths

Theory-Based Causal Induction

The ability to infer causal relationships from data is central to the growth of both scientific and everyday knowledge. Traditional explanations of human causal induction have tended to emphasize either domain-general covariation-based learning, or domain-specific knowledge about causal mechanisms. However, it is clear that both of these factors interact in most interesting cases of causal induction - the key questions are what prior knowledge is used, and how it interacts with statistical inference. I will present a computational framework that addresses these two questions, formulating the problem of causal induction as a Bayesian decision among a set of causal models generated by a domain theory. I will apply this framework to two phenomena of causal induction - inferring causal relationships from contingency data and learning the structure of physical systems.