Facilitation of repeated free operant discrimination reversals

Abstract

Two experiments are reported which studied the effects of several parameters on responses after discrimination reversals, “errors”, by retarded subjects. In Exp. 1 fixed-ratio and variable-ratio schedules of reinforcement and access to a switching key were investigated. In Exp. 2 the effects of a schedule change (FR 25 to FR 1) prior to reversal and a “count” of four reinforcements to reversal were studied with a design which permitted determining the effects of both variables simultaneously. Both experiments showed the diminution of errors after repeated reversals that has been reported in other studies of discrimination reversals and “learning set.” Access to a switching key resulted in reliably fewer errors in three of four subjects but many errors in excess of the ratios nevertheless were emitted. The number of responses after reversals may exceed FR schedules in direct relationship to the value of the FR—the higher the FR, the greater the number of responses in excess of the ratio. Schedule change prior to reversals and regularity in programming reversals resulted in errorless switching.