PUNISHMENT OF S-DELTA RESPONDING OF HUMANS IN CONDITIONAL MATCHING TO SAMPLE BY TIME-OUT.

Four neurotypical adults sat at a panel with three buttons and a screen. They learned a conditional matching task: if the sample color was red, press the left button when the comparison array appeared; if green, press the right.

Every wrong choice produced a 5-second blackout. The researchers then varied the time-out length (5 s vs 30 s) and how often it followed errors (every error vs every fifth error). They recorded accuracy across sessions.

Longer time-outs (30 s) cut errors by almost half compared with 5 s. Scheduling a time-out after every error (FR 1) worked better than after every fifth error (FR 5).

Accuracy climbed fastest when long and frequent time-outs were combined. The effect showed up within one session and stayed steady.

Ritchey et al. (2021) repeated the logic 60 years later on a touchscreen. Adults first earned points for swiping right, then that response was put on extinction. More minutes of reinforcement history caused bigger resurgence, confirming that dense early reinforcement strengthens later suppression.

Erickson et al. (2016) moved the idea into autism treatment rooms. Three children chose between a small immediate reinforcer and a larger delayed one. A mild delayed ‘No’ served as the punisher. Only one child showed clear sensitivity, reminding us that punishment parameters need individual probes.

Glodowski et al. (2020) used token boards with autistic adolescents. Token delivery every response (FR 1) sometimes suppressed responding, echoing the 1963 finding that denser consequences have more impact. Together these papers extend the lab result to clinical populations and new topographies.

If you run conditional-discrimination programs, tighten error correction. A 30-second break from tokens, music, or screen time beats a quick ‘try again.’ Deliver it every error, not every fifth. Probe the schedule if the learner has a long reinforcement history; resurgence may pop up later. Document response rate under any token or delay condition—individual sensitivity still rules.