FEED, FLEE, OR FACT-FIND: MODELING LATENT COGNITIVE-MECHANISTIC RESPONSES TO HETEROSPECIFIC ALARM CALLS FOR FORAGING CARDINALS

Heterospecific alarm calls allow animals to obtain information about predators from other species, but how receivers translate different forms of alarm information into defensive behavior remains poorly understood. In particular, signals may differ in whether they elicit rapid, relatively stereotyped responses or permit more graded evaluation of threat. We investigated this possibility by exposing foraging Northern cardinals (Cardinalis cardinalis) to playbacks of tufted titmouse (Baeolophus bicolor) calls associated with different predator contexts, including graded and immediate threats. To characterize responses, we developed a Bayesian latent-risk model that integrated three behaviors: feeding, vigilant looks, and departure, to estimate a shared dimension of perceived predation risk. Inferred risk increased from near-baseline levels under contact calls to intermediate levels under owl-associated call and the highest levels under ZZ calls. Increasing latent risk was associated with reduced expected seed intake and higher probability of complete feeding suppression. Vigilance showed a nonlinear relationship with inferred risk, increasing at intermediate levels before declining under the highest-risk treatment in favor of departure. This pattern is consistent with the possibility that receivers continue gathering information under intermediate threats but reduce vigilance when responding to more urgent danger. Although the experiment cannot fully establish whether these differences arise from canalized defensive responses or flexible cognitive evaluation, the results show how heterospecific alarm calls map onto distinct, coordinated behavioral responses.