Generalization of punishment effects-a case study.

One adult with intellectual disability received brief electric shocks when he hit himself. The shocks stopped the self-hitting while he sat in the treatment room. The team then watched to see if the behavior would stay low when he moved to other rooms or staff.

This was a single-case study. No other clients were treated.

The shocks worked fast. Self-injury dropped to zero during sessions. As soon as the client left the shock room, the hitting returned. The punishment did not carry over to the ward, the dayroom, or new staff.

The behavior was only suppressed where shocks had occurred. Therapeutic goals were not met because the change stayed trapped in one place.

Rincover et al. (1975) repeated the logic but used smelling salts instead of shock. They got the same result: punishment stopped self-injury only while the ammonia was present. Together, the two papers show the rule is about the contingency, not the tool.

Feinstein et al. (1988) moved the field forward. They first asked, "What is the behavior for?" then picked treatments matched to that function. Their matched plans outperformed arbitrary punishment, showing that function-based work later replaced the 1968 shock approach.

Wheatley et al. (1978) saw a similar failure in parent training. Brief demos did not generalize, just like brief shock did not. Both papers warn that narrow, one-setting interventions rarely travel.

If you use punishment today, plan for generalization from day one. Program common stimuli, train multiple caregivers, and fade the aversive once reinforcement systems take hold. Better yet, run a functional analysis first and build a treatment that addresses why the behavior happens. The 1968 case still matters because it shows what happens when we skip those steps.