Topographical variations in behavior during autoshaping, automaintenance, and omission training.

McIntire et al. (1987) watched pigeons during three classic procedures: autoshaping, automaintenance, and omission training. They filmed every head bob and step to map where the birds moved while the key light was on or off.

In omission training, pecking the key cancelled food. The team asked: would the birds still approach the light if pecking hurt their payoff?

Every bird kept walking toward the lit key and leaning in, even when pecking was blocked by the omission rule. Key pecks dropped, but partial pecks and head thrusts stayed.

During the dark inter-trial period the birds backed away. The light itself had become a powerful approach cue that could not be turned off by withholding food for pecks.

Vasconcelos et al. (2007) also filmed pigeons and found no extra work ethic—birds were indifferent to cues earned with high effort. Both labs show that fine-grained movement, not just response count, tells the real story.

Sheldon (1971) and FARMEMOORHEARSKELLEHER et al. (1964) chained schedules together and saw pausing shift with stimulus location. D’s work extends theirs: even when the final link is poisoned by omission, the earlier stimulus still pulls the bird in.

Together these papers warn us: once a stimulus becomes a conditioned reinforcer, simply blocking the terminal response may not erase the approach it controls.

If you run token boards or chained DTT, watch where the client moves, not just if they touch the target. A child may keep approaching the table even when the task no longer pays off. Build your extinction plan for the whole approach chain, not just the final response.