For video footage from past you can visit the individual event pages, or go to our YouTube Channel

To filter by event category, click on the event category link in the table below or use the menu on the right.

List of Past Events

The learning curve. What it really looks like and why it matters.

Dr. Charles R. Gallistel

Tuesday, November 18, 2003, 01:00pm - 02:00pm

Rutgers University, Center for Cognitive Science & Department of Psychology

Copy to My Calendar (iCal) Download as iCal file
 

November 18, 2003 at 1:00pm

Psychology Room 101, Busch Campus

Dr. Charles R. Gallistel

Rutgers University, Center for Cognitive Science & Department of Psychology

The learning curve. What it really looks like and why it matters.�

It is widely assumed that the development of learned behavior in standard animal learning paradigms is gradualr. This assumption is reinforced by the common practice of publishing learning curves obtained by averaging first across blocks of trials within each subject, and then across subjects. It has often been pointed out that this averaging may yield a curve that seriously misrepresents the progress of learning in individual subjects, but no one has determined what the learning curve looks like in individual subjects. I have analyzed data from different laboratories, using several different paradigms and species (pigeon autoshaping, rat autoshaping, mouse autoshaping, rat + maze, mouse water maze, and rat eyeblink). In all of these paradigms, conditioned behavior appears abruptly at close to its asymptotic strength. The transition from the pretraining baseline rate or strength of responding to the asymptotic level is more abrupt than is the rise of the psychometric functions in sensory threshold experiments (for example, the frequency of seeing curve). It is essentially a step. This finding bears on our conception of the nature of learning [Is it i) the strengthening of associative bonds, or, ii)� the gathering of information preparatory to making a decision to act?]� It also means that the learning curve cannot be used to estimate the rate of association formation (assuming, for the sake of argument, that there is such a process).

Dr. Charles R. Gallistel