Effects of Stimulus Sequence on P300 and Reaction Time in Schizophrenics

Subjects react to a random sequence of binary events as if the series were constrained by sequential rules. The reaction to a stimulus on trial n of a sequence is influenced by stimuli presented as far back as trial n 5 , as indicated by both choice reaction time and P300 measures. For example, subjects tend to respond more quickly to the continuation of a stimulus repetition than to its discontinuation (Kirby, 1976; Remington, 1969). These variations in reaction time and P300 amplitude have been attributed to trial-to-trial changes in the subject's expectancies. Our intent in this study was to determine whether or not patients diagnosed as schizophrenic formulate trial-to-trial expectancies for events in a random series and, if so, how their formulations compare to those of nonpsychiatric control subjects. The expectancy for a stimulus in a random series depends, for a constant level of global probability, on the operation of an exponentially decaying memory for the sequence of immediately preceding events (Audley, 1973; Squires et al., 1976). The P300 is a sensitive index of expectancy, with amplitude increasing monotonically as expectancy decreases (Duncan-Johnson and Donchin, 1977; Johnson and Donchin, 1980; Squires et al., 1976; Tueting et al., 1970). There is accumulating evidence that the P300 component is reduced in amplitude, but equal in latency, in schizophrenic patients compared to nonpsychiatric control subjects (Levit et al., 1973; Roth and Cannon, 1972; Roth et al.. 1980; Shagass et al., 1978; Shagass et al., 1977; Verleger and Cohen, 1978). A possible source of this reduction is a failure to generate trial-to-trial expectancies for events in a random series. This hypothesis is based, in part, on the finding that the sequence of stimuli preceding the eliciting event and the global probability of the event are additive determinants of P300 amplitude (Duncan-Johnson and Donchin, 1977; Squires et al.,