Initial evidence for probabilistic reasoning in a grey parrot (Psittacus erithacus).

Research has shown that some forms of inferential reasoning are likely widespread throughout the animal kingdom (e.g., exclusion, in which a subject infers the placement of a reward by eliminating potential alternative sites), but other types of inferential tasks have not been extensively tested. We examined whether a nonhuman might succeed in an experiment based on probabilistic reasoning, specifically, the ability to make inferences about a sample based on information about a population. A Grey parrot (Psittacus erithacus), previously trained to use English labels referentially to identify objects, observed a human researcher deposit 2 different types of items in a 3:1 ratio (e.g., 3 corks and 1 piece of paper) into an opaque bucket. One item was then randomly withdrawn while hidden from the parrot's view. When asked to identify the still-hidden object, the parrot's vocal responses tracked this 3:1 ratio over a large number of trials. Some levels of probabilistic reasoning therefore are not limited to humans, nonhuman primates, or even mammals. (PsycINFO Database Record

Human-Grey parrot (Psittacus erithacus) reciprocity: a follow-up study.

In a previous study (Péron et al. in Anim Cogn, doi: 10.1007/s10071-012.05640 , 2012), Grey parrots, working in dyads, took turns choosing one of four differently coloured cups with differing outcomes: empty (null, non-rewarding), selfish (keeping reward for oneself), share (sharing a divisible reward), or giving (donating reward to other). When the dyads involved three humans with different specific intentions (selfish, giving, or copying the bird's behaviour), birds' responses only tended towards consistency with human behaviour. Our dominant bird was willing to share a reward with a human who was willing to give up her reward, was selfish with the selfish human, and tended towards sharing with the copycat human; our subordinate bird tended slightly towards increased sharing with the generous human and selfishness with the selfish human, but did not clearly mirror the behaviour of the copycat. We theorized that the birds' inability to understand the copycat condition fully-that they could potentially maximize reward by choosing to share-was a consequence of their viewing the copycat's behaviour as erratic compared with the consistently selfish or giving humans and thus not realizing that they were indeed being mirrored. We suggested that copycat trials subsequently be performed as a separate experiment, without being contrasted with trials in which humans acted consistently, in order to determine if results might have differed. We have now performed that experiment, and shown that at least one Grey parrot--our dominant--responded in a manner suggesting that he deduced the appropriate contingencies.