The effect of negative stimulus presentations on observing-response rates.

Odom et al. (1986) worked with pigeons in a lab. They wanted to know if seeing an S- (a signal for no food) would make the birds look around more.

In three tests the team flashed the S- either after a peck or on its own. They counted how often the birds pecked a key just to see what color would come next.

Every time the S- showed up, the pigeons pecked the observing key more, even when the S- was not tied to their behavior.

The jump in pecks happened right after the S- appeared and then faded. The authors say this shows S- itself can drive looking, not act as a reinforcer.

Escobar et al. (2009) ran a similar pigeon test but lined up several S- cues late in extinction. They saw more observing only at the end of the chain, so they call S- a conditioned reinforcer under that timing. The two papers seem to clash, but the difference is in how the S- is scheduled: one long steady S- boosts looking without being a reinforcer, while a string of S- cues late in extinction can reinforce the look.

Bland et al. (2018) extended the idea by making S- contingent on a peck. The same kind of cue then acted as a punisher and cut responding. Together these studies show S- can have three faces: it can push looking, reinforce looking, or punish responding depending on when and how it is delivered.

Older work supports the backdrop: Shimp et al. (1974) showed pigeons love cues that signal food, while Azrin et al. (1969) found that even a cue for free food can briefly suppress work. Signals, good or bad, steer behavior more than we once thought.

When you run discrimination training, remember that flashing an S- can itself make a client look around more. If you want to test whether S- truly reinforces a skill, you must control for this automatic observing boost. Try inserting brief mask stimuli or separate trials so the S- does not contaminate your data.