Control by past and present stimuli depends on the discriminated reinforcer differential

Abstract

The extent to which a stimulus exerts control over behavior depends largely on its informativeness. However, when reinforcers have discriminative properties, they often exert less control over behavior than do other less reliable stimuli such as elapsed time. We investigated why less reliable cues in the present often overshadow stimulus control by more reliable cues presented in the recent past, by manipulating the reliability and duration of stimulus presentations. Five pigeons worked on a modified concurrent schedule in which the location of the response that produced the last reinforcer was a discriminative stimulus for the likely time and location of the next reinforcer. In some conditions, either the location of the previous reinforcer, or the location of the next reinforcer, was signaled by a red key light. This stimulus was either Brief, occurring for 10 s starting a fixed time after the most recent reinforcer, or Extended, being present at all times between food deliveries. Brief and Extended stimuli that signaled the same information had a similar effect on choice when they were present, but control by Brief stimuli weakened as time since stimulus offset elapsed. Control was divided among stimuli in the present and recent past according to the apparent reliability of the information signaled about the next reinforcer. More reliable stimuli in the present degraded, but did not erase, control by less reliable stimuli presented in the recent past. Thus, we conclude that less reliable stimuli in the present control behavior to a greater degree than do more reliable stimuli in the recent past because these more reliable stimuli are forgotten, and hence their relation to the likely availability of food cannot be discriminated.

Stimulus control depends on the subjective value of the outcome

“…Our findings contribute to a body of research that demonstrate that an organism's own behavior is an effective discriminative stimulus when environmental conditions permit enduring control (e.g., Cowie et al, 2011Cowie et al, , 2017Davison, 2018;Fox & Kyonka, 2013Killeen, 1978;Kramer & Rilling, 1970;Zentall et al, 2001;Zentall et al, 2008). The keypeck that produced entry into the terminal link exerted prolonged control over terminallink response rates when initial link stimuli were Conflicting and when they were not (Signaled and Conflicting probe trials, Fig.…”

“…These effects were particularly apparent for stimuli signaling the Larger-Later (typically less-preferred) outcome. When stimuli and the outcomes they predict are separated in time (e.g., see Cowie et al, 2017, for a discussion), detection of the correlation depends on whether the first event is remembered at the time the second event occurs. If the stimulus is sometimes forgotten, its correlation with the outcome will appear less reliable in the sense that the outcome will sometimes appear to have occurred in the absence of the stimulus (e.g., see Cowie et al, 2017).…”

“…Control over behavior by any brief stimulus will be subject to the degree to which the meaning of that stimulus may be brought into the present. When organisms must remember the specific event, control is often transient (see Cowie et al, 2017;Zentall et al, 2001;Zentall et al, 2008). Conversely, stimuli that remain present in some form-for example, keylight colors (e.g., Cowie et al, 2011;, and enduring differences in body positioning-serve the same discriminative function as the entry-response itself, effectively bringing the past into the present (e.g., see McMillan & Roberts, 2012;Rayburn-Reeves et al, 2018).…”

“…there is a delay between a brief stimulus and subsequent reinforcers (e.g., Cowie et al, 2017); when highly similar stimuli correlate with different reinforcer conditions (e.g., Alsop & Davison, 1991;Miller et al, 1980;Miranda-Dukoski et al, 2014); and when irrelevant events occur between the relevant stimulus and subsequent reinforcers (e.g., Davison, 2018). All these different manipulations impede discrimination of some element of a correlation, or of the link between elements.…”

Being there on time: Reinforcer effects on timing and locating

“…Correlations that create environmental signposts are necessarily extended in time: Milliseconds pass between the presentation of a conditioned stimulus and an unconditioned stimulus (e.g., Pavlov, ); seconds pass between one fixed‐interval‐schedule reinforcer and the next (e.g., Ferster & Skinner, ; Schneider, ); hours pass between the availability of food that may be cached and the occurrence of breakfast the next day (Raby, Alexis, Dickinson, & Clayton ). The ability of a signpost event to exert stimulus control over behavior depends in part on the organism's ability to remember an event across the relevant timescale (e.g., see Cowie, Davison & Elliffe, ). Correlations cannot be learned when one event is forgotten before the next event is encountered.…”