Prediction outcome and choice reaction time: A memory-dependent relationship.

Abstract

In two-choice reaction time (RT) situations, the interval between the prediction and the presentation of successive stimuli was either 3, 7, or 11 sec. The 20 5s of one group (Group Stim) pressed a left-hand trigger following one stimulus and a right-hand trigger following the alternative stimulus; 5s of another group (Group Pred) identified the prediction outcome on each of the 200 trials by making one response if the prediction was correct and the other response following an incorrect prediction. The influence of the prediction-stimulus interval implicated both expectancy and memory factors in choice RT. That is, RTs were faster following correct than incorrect predictions, and this effect of prediction outcome was greatest following the shortest interval. The reliable effects of prediction outcome on RT were more prominent for Group Pred than for Group Stim.When required to predict the sequential occurrence of stimulus alternatives in a choice reaction time (RT) paradigm, 5s react markedly faster to correctly predicted than to incorrectly predicted events (e.g., Geller & Pitz, 1970;Geller, Whitman, & Karris, 1972;Keele, 1969). The present experiment was designed to determine if the temporal interval occurring between the prediction and the presentation of a stimulus would influence the effects of the prediction outcome on reaction latency. The three prediction-stimulus intervals used (3, 7, and 11 sec.) were within a range in which short-term memory (STM) for consonant trigrams was shown to decay (Peterson & Peterson, 1959). Hence, an STM model might hypothesize that the prediction outcome effect would be most apparent at the shortest interval (i.e., 3 sec.), since at this interval the memory trace of a verbalized expectancy would be strongest. Following the longest interval (i.e., 11 sec.) the effect of prediction outcome on choice RT should be least prominent, since the verbal STM for the stimulus 1 The authors thank David A. Williams for his assistance in collecting the data and Western Electric Company, Inc., for its donation of electronic supplies.2 Requests for reprints should be sent to E.

The time course of the verbal prediction effect

“…The assumption implicit in either proeedure is that, onee a verbal prediction has been uttered, it will remain in memory, unchanged, until the stimulus in presented. Geller and Whitman (1972) have ehallenged this assumption by varying the time between a subject's predietion and the stimulus from 3 to 11 sec. Their results showed that, as the PSI inereased, the differenee in RT between eorreetly and ineorreetly predicted stimuli decreased, that is, aPrediction Outeome by PSI interaction.…”

Studying short-term storage of expectancies with reaction time

“…The apparatus and general procedure were similar to those used by Geller and Whitman (1972). The order of events on each of 198 RT trials were as follows: (1) The experimenter prompted the subject's stimulus prediction by illuminating the center, horizontal line (1.3 cm) of the seven-segment stimulus readout; (2) the subject predicted which of the two possible stimulus events would occur, a U symbol (called up) or a n symbol (called down); (3) with the left forefinger, the subject depressed one of two spring-return levers to indicate a stimulus prediction (these prediction levers were located on a panel to the left of the subject and were separated on a vertical axis by 7.7 cm; the uppermost lever was used to designate a prediction of U); (4) the subject's depression of the prediction lever initiated a time interval of 3, 7, or 11 sec; (5) a .25 -sec "ready" buzzer occurred and initiated a fixed interval of 1.5 sec; (6) the U or n symbol (5 x 1.3 cm) was presented by illuminating the appropriate segments of the seven-segment stimulus readout; (7) the subj·ect identified the stimulus presentation as quickly as possible by pulling a left-or right-hand reaction trigger.…”

“…Supporting this simple expectancy interpretation of the PO effect are the observed relationships between choice RT and factors that presumably qualify the subject's stimulus predictions. For example, choice RT to correctly predicted stimuli was an inverse function of the confidence associated with the prediction , the probability of a correct prediction (Geller, 1974;, and the relative frequency of the stimulus predicted (e.g., Geller & Pitz, 1970; Geller, Whitman, Wrenn,& Shipley, 1971).The present research was designed to follow up an observation that the PO effect decreased as the interval between th~ subject's prediction and the stimulus presentation mcreased (Geller & Whitman, 1972). In the Geller and Whitman study, the prediction-to-stimulus (P-S) interval was 3, 7, or 11 sec, varying randomly among the 198 trials for each subject.…”

“…Thus, the decrease in the PO effect as a function of the P-S interval was presumed to reflect a decay in the memory trace of the subject's verbalized expectancy over the range of intervals used. A prominent weakness in the Geller and Whitman (1972) study was the lack of control over the subjects' cognitive activities intervening between their predictions and the stimulus presentations. For example, after making a prediction, subjects could rehearse their prediction (thus strengthening their expectation) or think about task-irrelevant items (thus weakening their expectancy); therefore, it was actually inaccurate for Geller and Whitman to infer expectancy decay rather than expectancy interference from their findings .…”

“…The present research was designed to follow up an observation that the PO effect decreased as the interval between th~ subject's prediction and the stimulus presentation mcreased (Geller & Whitman, 1972). In the Geller and Whitman study, the prediction-to-stimulus (P-S) interval was 3, 7, or 11 sec, varying randomly among the 198 trials for each subject.…”