Abstract
Subjects interacted with a confederate in video taped dyads and were later asked to: (a) recognize whether certain specific communication behaviors had occurred during the conversation and (b) estimate the frequency with which certain behaviors occurred in it. Subjects consulted memory for conversational behavior more than they relied upon implicit theories to provide verbal reports. Subjects were better at recognizing the gist of a remark than its verbatim content. As Ericsson and Simon's theory predicts, subjects were better able to recognize verbal than nonverbal behaviors (when elicited in verbal form) and were better able to recognize specific behaviors than to generate frequency counts. Subjects were better at recognizing their own than other's behavior.