The failure of stimulus control after presence-absence discrimination of click-rate.

Researchers worked with pigeons to test two ways of teaching a discrimination.

One group learned fast-versus-slow clicks. The other group only heard clicks during reinforced trials and silence during non-reinforced trials.

After training, both groups saw a range of click speeds to see which birds pecked only at the trained rate.

Birds trained with fast-versus-slow clicks pecked mostly at the trained speed. Their generalization curve looked like a steep hill.

Birds trained with clicks-present versus clicks-absent pecked at almost every speed. Their curve was flat.

Presence-absence training failed to create tight stimulus control.

Wildemann et al. (1973) ran a near-copy of this study and got the same sigmoid curves, showing the result is reliable.

Mello (1966) saw the same flat gradients when pigeons learned punishment without a second stimulus. Together the papers say: contrast two values on one dimension, or control stays loose.

Lyons (1995) later bundled these findings into a tutorial, telling trainers to avoid simple presence-absence drills if they want crisp stimulus control.

When you teach a learner to tell 'red' from 'not red' you may get sloppy responding. Swap in 'red versus green' and control tightens. Check generalization early; if the gradient is flat, add an intradimensional contrast instead of more trials alone.