Response Strength versus All-or-None Interpretation of Reponse Occurrence

Abstract

The data to be presented involve the first four trials in the learning of a discrimination among two non-overlapping patterns of eight lights each, selecred from an array of 24 lights. The procedural details have been described elsewhere (Schoeffler, 1954), and only those points germane to the present discussion are presented here. Discrimination learning.-Two hundred and thirty-four Ss were trained to discriminate between two patterns of lights (SI and SII). In order to accomplish this, a pattern was presented and S moved a lever right or left. After the response had been made, S was informed which response was correct on that trial. The correct response to stimulus SI will be called R~I and the correct response to SII will be called Rsll. The stimulus patterns were presented in a random sequence with the restriction that each block of 10 trials contained 5 SIs and 5 SIIs. The analysis of the data concerns the probability that an S making a correct response on a particular trial will make anochet correct response on the next trial. The data on each trial refer to the correctness or incorrectness of the response of each S. RESULTS AND DISCUSSION Table 1 indicates the number of Ss who responded correctly and the number who responded incorrectly on Trial 1. The correct and the incorrect responders on the first trial are then each divided into 2 groups depending on whether they responded correctly or incorrectly on the second trial. In the same way these 4 groups are then further divided into 2 groups each, depending on the correctness of the response on the third trial, yielding 8 groups, and these 8 groups are subdivided to yield 16 groups on the fourth trial each having a different sequence of correct and incorrect responses. These data are entered in the column labeled Observed. The over-all proportion of correct responding on each trial is indicated at the bottom of the table. The concept of response strength is interpreted in the present context as follows. In this experiment, two (and only two) responses are available to S, vzz., RsI and RsI~. The proportion of a group producing a correct response when a given stimulus is presented, is interpreted as the probability for each individual that he has learned the proper response to a stimulus, and chis