Minimizing predation risk, especially for young or naïve individuals, can be achieved by learning to recognize predators. Embryonic learning may optimize survival by allowing for the earliest possible response to predation threats posthatch. However, predatory threats often change over an individual’s lifetime, and using old information can be detrimental if it becomes outdated. Adaptive forgetting allows an individual to discount obsolete information in decision-making and instead emphasize newer, more relevant information when responding to predation threats. Little is known about the extent to which young individuals can learn and forget information about predation threats. Here we demonstrate that rainbow trout 1) are capable of learning from both conspecific and heterospecific alarm cues as embryos, newly hatched larvae, and free-swimming larvae, 2) exhibit adaptive forgetting of predator information at all stages, and 3) display dynamic adaptive forgetting based on the ontogeny of learning. Specifically, fish that learned information as embryos retained the information for longer periods than those that learned the same information as newly hatched alevins.

When faced with a changing environment, some species appear to adapt quickly, while others seem unable to update the value of environmental cues on which they base their decisions, leading them to display seemingly maladaptive responses. While behavioural and cognitive plasticity are two traits that should predict the ability of species to update the value of environmental cues, we argue that this flexibility may be constrained by ontogeny. While sensitive periods have been shown to exist for establishing an individual’s food, habitat and mate preference, no studies have established the existence of a cognitive sensitive period for predation‐related information. In this study, we used wood frogs, Lithobates sylvaticus, to demonstrate the existence of a sensitive period for predation‐related information, with risk information learned as embryos maintained for more than 5 weeks, while the same information learned as tadpoles was unused after just 10 days. Next, we demonstrated that tadpoles that had learned a cue as safe as embryos were unable to update the cue as risky after three fear conditioning attempts, while tadpoles that learned the cue as safe a few days prior did successfully update the cue as risky after three conditionings. We coined the term “cognitive resonance” to describe how information learned early in life can have marked cognitive consequence later in life, affecting not only the duration for which information learned is actively used in decision‐making, but how this information can interfere with the acquisition of up‐to‐date information about the environment. Cognitive resonance might be beneficial in stable environments where the change in the value of a cue is relatively small through time, but it can quickly become costly in environments where the identity of potential threats changes quickly, as in the case of introduced species. A plain language summary is available for this article.

No abstract

Organisms are exposed to a wealth of chemical information during their development. Some of these chemical cues indicate present or future dangers, such as the presence of predators that feed on either the developing embryos or their nearby parents. Organisms may use this information to modify their morphology or life-history, including hatching timing, or may retain information about risk until it gains relevance. Previous research has shown predation-induced alterations in hatching among embryonic minnows that were exposed to mechanical-injury-released alarm cues from conspecific embryos. Here, we test whether minnows likewise hatch early in response to alarm cues from injured adult conspecifics. We know that embryonic minnows can detect adult alarm cues and use them to facilitate learned recognition of predators; however, it is unknown whether these adult alarm cues will also induce a change in hatching time. Early hatching may allow animals to rapidly disperse away from potential predators, but late hatching may allow animals to grow and develop structures that allow them to effectively escape when they do hatch. Here, we found here that unlike embryonic fathead minnows (Pimephales promelas) exposed to embryonic cues, embryonic minnows exposed to adult alarm cues do not exhibit early hatching. The ability of embryos to recognize adult alarm cues as a future threat, but not a current one, demonstrates sophisticated ontogenetic specificity in the hatching response of embryonic minnows.

Many aquatic prey animals release chemical cues upon being captured by a predator. These chemical cues, referred to as alarm cues, may act to warn nearby individuals of danger. For the cues to be useful, fish must be able to discern if they are indicative of a real threat; cues from conspecifics in different age groups may be irrelevant due to size- and habitat-related shifts in predation risk. We test the response of newly-hatched rainbow trout, Oncorhynchus mykiss, to three concentrations of alarm cues from conspecifics from two age groups: newly-hatched versus six-month-old juveniles. Newly-hatched trout demonstrated a significant fright response to all three concentrations of alarm cues, but showed no difference in strength of response based on either concentration or age of the cue donor. We propose that the newly-hatched trout did not respond differently because of the high risk of predation that they face during this life stage.