Pigeons' preference for free choice: number of keys versus key area.
Two small choices beat one big option—separate response keys are themselves reinforcing.
01Research in Context
What this study did
The team placed pigeons in a chamber with two white keys.
In one condition the birds could peck either key.
In the other condition only one key worked; the second key stayed dark.
Total key size stayed the same across both set-ups.
Birds earned grain on a concurrent-chains schedule.
What they found
The pigeons picked the two-key set almost every time.
They wanted separate choices, not a bigger target.
Even when the single key was twice as wide, they still chose the pair.
How this fits with other research
Yuwiler et al. (1992) and Macdonald et al. (1973) showed pigeons follow the matching law on two-key schedules.
Fabbretti et al. (1997) kept the same law in place but asked why the birds like two keys.
Pisacreta (1982) seems to clash: his pigeons stuck to one rigid pattern even with many options.
The difference is procedure.
R gave 720,000 equal paths; no matter which path the bird took, grain came.
D kept the payoff tied to the key.
With stakes still linked to each key, variety stayed valuable.
Why it matters
When you give learners two clear buttons, cards, or tasks, you raise reinforcement value without adding extra payoff.
Try splitting one large picture card into two smaller ones during mand training.
The learner still gets the same item, but the act of choosing adds free reinforcement for the response.
Want CEUs on This Topic?
The ABA Clubhouse has 60+ free CEUs — live every Wednesday. Ethics, supervision & clinical topics.
Join Free →Cut one large 4-inch PECS card into two 2-inch cards and let the learner pick.
02At a glance
03Original abstract
In concurrent-chains schedules, pigeons prefer terminal links that provide two keys correlated with reinforcers (free choice) over those that provide only one key (forced choice), terminal-link reinforcement rates being equal. With same-size keys, free choice provides a larger area available for pecking. Preferences were examined using terminal links that differed in key number only (one or two) or key size only (small and medium or medium and large), or that equated the area of the two free-choice keys with that of the forced-choice key. Medium (standard) keys were typically preferred to small keys, but indifference was typically obtained between medium and large keys. The size preference usually overrode free-choice preference with one medium key pitted against two small keys, but free-choice preference was reliably observed with one large key pitted against two medium keys. In other words, preferences were a joint function of key number and key area, implying that free-choice preference is not reducible to preference for larger key areas. Free-choice preference requires separate keys rather than larger areas; the relevant behavioral units are the discriminated operants correlated with each terminal-link key rather than classes defined by topographical features such as area or perimeter.
Journal of the experimental analysis of behavior, 1997 · doi:10.1901/jeab.1997.68-349