Transfer of matching-to-figure samples in the pigeon.

Sprague et al. (1984) worked with three pigeons in a lab.

The birds learned to match figure samples using six keys.

Only some comparison keys lit up at first, then more appeared step by step.

After training, the team showed brand-new figures to see if the birds could still match.

All three pigeons matched the new figures correctly.

The birds acted as if they had learned a rule: "pick the one that looks like the sample."

No extra training was needed for the novel shapes.

Cohen (1969) saw the same thing earlier with adjustable comparisons, so the new study extends that work to a six-key setup.

Haemmerlie (1983) and Howard (1979) also found transfer in pigeons, but with locations or colors instead of figures. Together they show birds can abstract rules across different kinds of stimuli.

Saunders et al. (1988) looks like a contradiction at first glance. Those birds failed symmetry and transitivity tests after standard matching-to-sample. The difference is the goal: the 1984 paper only asked "can they match new figures?" while the 1988 paper asked "do the birds show true equivalence relations?" Matching new shapes does not guarantee full equivalence class formation.

For BCBAs, the lesson is to check what kind of transfer you need. If you want a child to match new letters, incremental matching-to-sample may be enough. If you need full equivalence—reading, spelling, and saying the letter—you must test symmetry and transitivity later. Start with simple transfer, then probe for the bigger concept.