The Department of Psychology Distinguished Scholar Lecture Series is a three-day public lecture series sponsored by the Department of Psychology, Faculty of Arts and Faculty of Science. Each year the Department invites a renowned psychologist to present three one-hour lectures. The event is normally scheduled in the fall.
Distinguished Scholar Lecture Series 2017
The 39th Annual Distinguished Scholar Lecture Series spanned from January 11 to 13, 2017. This year's guest speaker was Dr. William H. Warren, Chancellor's Professor at Brown University. Dr. Warren engaged The Dynamics of Perception and Action through three seminars focused on Locomotion, Crowd Behaviour, and Navigation respectively. The seminar series was video-recorded and can be watched here.
39th Annual Distinguished Scholar Lecture Series
Dr. William H. Warren - The Dynamics of Perception & Action
January 11 - 13, 2017 | 3 - 4:30 PM
Lecture I : Locomotion
Jan 11 | CCIS 1-140
How do humans and other animals locomote through a complex, changing environment? I will argue that paths of locomotion emerge in an on-line manner from the interaction between an agent and its local environment. Based on studies of visually-controlled human walking in virtual environments, we create simple dynamical models of steering to a goal, obstacle avoidance, interception, and moving-obstacle avoidance. By combining these elementary behaviors, our ‘pedestrian model’ can predict locomotor trajectories in more complex environments. Some strategies are strikingly similar to those observed in insect flight control, suggesting very general principles. The results demonstrate that locomotor behavior can emerge on-line as a stable solution of the system’s dynamics, without appealing to an internal world model or explicit path planning.
Lecture II : Crowd Behaviour
Jan 12 | CCIS 1-140
What accounts for patterns of collective motion, such as bird flocks, fish schools, and human crowds? Such collective behavior is thought to emerge from local interactions between individuals. The key to the problem is thus to understand the local ‘rules’ or visual coupling that govern these interactions. There are many such models of collective motion, but precious little empirical evidence. Based on human experiments with virtual crowds, we model ‘following’ and characterize the neighborhood of interaction in a crowd. We than use multi-agent simulations of the pedestrian model to predict crowd behavior, and compare the results with motion-capture data on real human crowds. Scenarios like Grand Central Station, Human Swarm, and Counterflow can be successfully simulated with a few components of the pedestrian model. The results support the view that crowd dynamics emerge from local interactions, consistent with principles of self-organization.
Lecture III : Navigation
Jan 13 | CCIS L1-140
How do humans and other animals navigate to places beyond the sensory horizon, which are ‘off-line’? It is often assumed that spatial navigation implicates a ‘cognitive map,’ an internal representation of the environment with a Euclidean geometric structure. Such spatial knowledge could be built up from path integration, as hypothesized for the grid and place cell system. Our experiments on navigation in virtual environments converge with previous research to show that humans (i) have poor and discontinuous path integration, (ii) rely heavily on visual landmarks, and (iii) take highly unreliable shortcuts that (iv) violate the metric postulates. The results imply that humans do not build a geometrically consistent map. I will suggest that spatial knowledge is better characterized as a labeled topological graph that incorporates rough local metric information from the path integration system. Graph knowledge could constrain the pedestrian model by specifying the approximate direction to unseen goals, while the locomotor path emerges on-line. Apparently Euclidean behavior may thus result from minimal spatial knowledge together with on-line control of locomotion.