Inverse relation between choice and local response rate with a schedule of response-produced blackouts.

Arnett (1972) let three pigeons pick between two keys in a concurrent-chain setup. One key led to a schedule where quick pecks turned the lights off for half a second. The other key led to the same food schedule with no blackouts.

The blackout was meant to bump up the local peck rate. The question: would the birds prefer the key that made them peck faster?

Two birds avoided the key that gave blackouts even though it made them peck faster. One bird did not care and picked both keys equally.

So, higher local response rate did not equal higher preference. The blackout side lost appeal for most subjects.

Millard (1979) later showed that the same blackout trick beats the usual change-over delay at stopping superstitious switching. Arnett (1972) showed the cost: blackout can also cut preference for that side.

Cohen et al. (1990) saw similar mixed reversals when long signaled delays were added. Both studies warn that local schedule tweaks can flip choice in ways you might not predict.

Ploog (2001) found that bigger food almost always wins. Arnett (1972) reminds us that non-food schedule events like blackout can still override that pull for some learners.

When you add brief timeouts or response-cost flashes to boost rate, watch what the learner actually chooses next. Faster responding may come at the price of lower preference for that task or area. If you see avoidance, try thinning the blackout or pairing it with extra praise or tokens so the schedule stays attractive while still shaping brisk responding.