A Theoretical Approach to the Late Components of the Event-Related Brain Potential

In the mid seventies, Brigitte Rockstroh and myself studied avoidance learning in human subjects. We presumed that brain potentials might be a better predictor of resistance to extinction than, for example, autonomic responses. In one experiment, subjects could escape a highly aversive noise by pressing a button fast enough. As it happens in experimental laboratories, a loose connection in the wire which was attached to the button, only randomly transmitted the subject's response to the computer. Consequently the subject experienced a sudden V unexpected change in response contingency, which, of course, we completely denied when reported to us. The complaint about the frequent and unpredictable punishment was, at first, answered only with a request to follow adequately the instruction. When we examined the subject's EEG responses we detected a marked negative wave following many of the single motor responses. This was unusual as generally a positive wave accompanies the termination of an overt response. Systematic studies soon revealed that a pronounced negativity develops, whenever expectations of relevant predictions are not fulfilled, whenever there is a need for ongoing processing (Rockstroh et al. 1979; Elbert et al. 1982; Elbert 1986). Controlled experiments soon revealed that such slow negative variations of the EEG-baseline had little to do with the physical properties of the stimulus as such, but were determined by the psychological context the stimuli were put in. Kutas and Hillyard (1980, 1984) extracted a similar finding from an elegant series