The Dead Man Test: a Preliminary Experimental Analysis

Critchfield and his team tested the Dead Man Test. This rule says, "If a dead man can do it, it isn’t behavior."

They watched three deceased people for any sign of movement or sound. The team used a simple checklist and video to be sure.

The study is tiny. It is the first time anyone tried to check the rule with real data.

No behavior showed up. The dead men stayed still and quiet, just as the rule predicts.

The authors warn that three cases are not enough to prove the rule. They call the work a first step, not a final answer.

Sarber et al. (1983) tried something close. They gave music to coma patients and saw small responses. Their work shows that even tiny behavior can be found if you look hard. Critchfield’s study flips the question: can anything be seen when life is gone? Both use tight single-case watching, but one hunts for the smallest sign while the other checks for total silence.

Fraley (1998) wrote that death is just another event in a behavior-analytic view. Critchfield gives that idea a quick lab test. The 1998 paper talks theory; the 2018 paper adds a pinch of data.

Mottron (2015) warns that some big ideas in assessment lack proof. Critchfield’s work echoes the same worry. The Dead Man Test is taught every day, yet this is the first attempt to check it.

You use the Dead Man Test to decide if a goal is really behavior. This study says the rule held in three cases, but we still lack solid proof. Keep using the test, yet stay open to better data. When you write goals, pair the rule with clear, measurable action words so no one has to guess what counts.

The dead man's test is a rule of thumb attributed to Ogden Lindsley, who proposed it in 1965 as part of his precision teaching work: if a dead man can do it, it is not behavior, and if a dead man cannot do it, it is behavior. The test exists to catch a specific error in how we define targets, which is describing the absence of action as if it were an action.

The logic is simple. Behavior analysis measures what organisms do. A corpse can 'sit quietly,' 'refrain from hitting,' and 'stay in its seat' flawlessly, forever. If a goal can be met by doing nothing at all, then it does not describe behavior, and anything built on it, including measurement, reinforcement contingencies, and progress reports, inherits the problem.

The test is a screening tool, not a law of nature. It gets invoked constantly in coursework and supervision precisely because dead-man targets are so easy to write by accident, especially when the referral concern is framed as making a behavior stop.

Fails the test (a dead man can do it): not hitting peers, sitting quietly with hands still, staying in the designated area, refraining from screaming, being off task zero percent of the interval. Each one describes an absence. You cannot deliver a reinforcer contingent on nothing happening and expect a specific repertoire to grow.

Passes the test (a dead man cannot do it): raising a hand, walking to the schedule board, saying 'help please,' writing an answer, placing hands on the desk and manipulating materials, handing over a break card. Each one is an observable, countable action with a beginning and an end.

The repair is usually mechanical: translate every dead-man goal into the active behavior you want instead. 'Not hitting' becomes 'requests a break when work is presented.' 'Sitting quietly' becomes 'completes worksheet items' or 'looks at the speaker and answers questions.' This is the same logic that pushes intervention toward differential reinforcement of alternative or incompatible behavior rather than extinction alone.

Dead-man targets corrupt measurement. An interval recording of 'no aggression' scores a perfect session for a student who slept through it, and a token system that pays for stillness teaches nothing you could graph as an acquisition curve. Passing your operational definitions through the dead man's test is one of the fastest quality checks a BCBA can run on a behavior plan.

This paper is behavior analysis having fun with its own axioms, and that is the point. The authors note that the assumption behind the test, that behavior is absent in the deceased, had never been checked empirically, so they observed three reasonably deceased individuals under three conditions and detected no behavior. The result is consistent with the rule, and the deadpan method is a reminder that even foundational assumptions can be stated as testable predictions.