Bringing the past into the present: Control by exteroceptive stimuli and key-peck location in a concurrent-chains procedure
Introduction
Behavior depends not only on discriminative stimuli in the present, but also on stimuli that occurred in the recent past (e.g., Cowie et al., 2017, Cowie et al., 2020). Control by past stimuli may be adaptive, as when a previously presented stimulus signals the likely location of future reinforcers (e.g., Cowie et al., 2017; Santos et al., 2019), or may impede discriminative performance, as when control by past, now-irrelevant, stimuli compete with control by more recent, relevant stimuli (e.g., Cowie et al., 2017; Edhouse and White, 1988; Gomes-Ng et al., 2018).
Past stimuli control present behavior to the extent that a specific stimulus occurrence can be remembered by the organism – an ability that depends on the organism’s memory systems (e.g., Lind et al., 2015) and on stimulus characteristics. Some reliable stimuli that are separated in time from an outcome fail to exert complete control over behavior (e.g., Cowie et al., 2017, Cowie et al., 2020), while others exert enduring control (Cowie et al., 2020, Gomes-Ng et al., 2019a, Urcuioli, 1984, Urcuioli, 1985, Weaver et al., 1999). This difference may lie in a stimulus’ ability to create the occasion for other cues that ‘bridge the gap’ between the stimulus and reinforcer delivery. For example, when there is a delay between a stimulus and subsequent discrimination (as in delayed matching-to-sample), differential sample behavior (e.g., body positioning) may exert control, sometimes even overshadowing control by the discriminative stimulus itself (e.g., Gomes-Ng et al., 2019a; Urcuioli, 1984, Urcuioli, 1985; Urcuioli and Honig, 1980; Weaver et al., 1999).
Cowie et al. (2020) recently showed that control by pigeons’ key-peck location overshadowed control by past discriminative stimuli in a concurrent-chains procedure in which one alternative led to a terminal link ending in a smaller reinforcer delivered after an 8-s delay, and the other to a link ending in a larger reinforcer after a 48-s delay. In some trials, color-alternating stimuli (hereafter, termed signals) during initial links differentially signaled the trial outcome, or conflicting outcomes (i.e., smaller-sooner and larger-later). During initial links, the signaled key was never chosen exclusively, indicating imperfect signal control, and subjects were biased towards the smaller-sooner alternative, indicating outcome control (i.e., control by the outcomes of each alternative). In terminal links, response patterns were similar in Signaled and Unsignaled trials. This suggests that responding was controlled by the key peck producing terminal-link entry, as opposed to the signals. Such key-peck control was also evident in Conflicting trials.
Why did control by key-peck location overshadow signal control in Cowie et al. (2020)? One possibility is that signals exerted weak control because they were temporally separated from trial outcomes. Although key-peck location was also separated from trial outcomes, enduring differences in body positioning during terminal links may have bridged the temporal gap, resulting in key-peck control. Presently, it is unclear to what extent the temporal separation of signals from trial outcomes contributed to Cowie et al.’s findings. Would control by signals have been stronger, and control by trial outcomes and key-peck location weaker, if signals were temporally closer to trial outcomes? The present experiment addressed this question. We replicated Cowie et al.’s experiment, except that signals remained present during terminal links, thus removing the temporal separation between signals and trial outcomes. If Cowie et al.’s results were caused by the transience of the signals, then control by such signals should be stronger in this experiment. In contrast, if Cowie et al.’s results were related to some other factor, such as outcome control, then they should be largely replicated here.
Section snippets
Subjects and apparatus
Five adult Columba livia pigeons (sexes and ages unknown), numbered 112–116, served as subjects. Subjects were maintained at 85% ± 15 g of free-feeding body weight. All had previous experience in a concurrent-chains procedure (Cowie et al., 2020). Pigeons were housed individually in custom metal cages (375 m high × 375 mm deep × 370 mm wide), containing two elevated wooden perches. The right wall of each cage contained an operant panel with three 20-mm diameter response keys centered 100 mm
Results
We analyzed data from the last ten sessions of testing (termed Condition 3 in Cowie et al., 2020) to ensure that comparisons between trial types used data collected in the same sessions. In the analyses reported here, we also present relevant data from Cowie et al. (2020) for comparison (Past-Signal condition).
Discussion
We investigated whether stimuli that signaled future outcomes in a concurrent-chains procedure (Fig. 1) exerted stronger control when those signals were present until the outcome (Present-Signal condition), than when they occurred sometime before the outcome (Past-Signal condition; Cowie et al., 2020). During initial links, signal control was stronger in the Present-Signal condition than the Past-Signal condition. During terminal links, responding appeared to be controlled strongly by the
CRediT authorship contribution statement
Stephanie Gomes-Ng: Conceptualization, Methodology, Software, Formal Analysis, Investigation, Writing – Original Draft, Writing – Review & Editing, Visualization, Project Administration, John Y. H. Bai: Conceptualization, Methodology, Software, Investigation, Writing – Review & Editing, Supervision, Jason Landon: Conceptualization, Methodology, Writing – Review & Editing, Sarah Cowie: Conceptualization, Methodology, Resources, Writing - Review & Editing, Supervision.
Author Note
The authors thank the members of the Experimental Analysis of Behaviour Research Group at The University of Auckland for their help in running the experiment, and Mike Owens who looked after the pigeons. This experiment was conducted under Approval 001396, granted by The University of Auckland Animal Ethics Committee.
References (29)
- et al.
Environment tracking and signal following in a reinforcer-ratio reversal procedure
Behav. Process.
(2018) - et al.
Animal memory: a review of delayed matching-to-sample data
Behav. Process.
(2015) - et al.
What makes information value: signal reliability and environmental uncertainty
Anim. Behav.
(2006) - et al.
Dynamic cue use in pigeon mid-session reversal
Behav. Process.
(2017) Reinforcement schedules and psychophysical judgements: a study of some temporal properties of behavior
- et al.
Reinforcement: Food signals the time and location of future food
J. Exp. Anal. Behav.
(2011) - et al.
Does overall reinforcer rate affect discrimination of time-based contingencies?
J. Exp. Anal. Behav.
(2016) - et al.
Control by past and present stimuli depends on the discriminated reinforcer differential
J. Exp. Anal. Behav.
(2017) - et al.
Stimulus control depends on the subjective value of the outcome
J. Exp. Anal. Behav.
(2020) Divided stimulus control: Which key did you peck, or what color was it?
J. Exp. Anal. Behav.
(2018)