The definition of learning has been the subject of long, unresolved debate, and no definition is universally accepted. However, most students of animal learning agree that learning involves a repeatable change in behavior with experience that persists for some time after experience ends (i.e., there is evidence of memory). Furthermore, with a few important exceptions, learned behavior changes gradually with continued experience to some asymptote; wanes if not continually reinforced; can often be undone by a new type of experience; and is more suited to the environment in some way (i.e., associated with higher fitness) than before learning took place.
Learning is just one form of behavioral plasticity. Learning is sometimes referred to as a form of phenotypic plasticity in which the phenotype is behavior. This is not imprecise, but can be somewhat misleading because behavior is itself a form of phenotypic plasticity. Behavior, like any form of phenotypic plasticity, can be described by a 'reaction norm' that relates an animal's phenotype, in this case a set of morphologies generated by motor outputs, to particular environmental states. Learning in turn can be described as a mechanism by which the reaction norm representing a particular behavior is modified by experience with the environment. Learning is thus a mechanism of plasticity in behavior, which is itself a type of plasticity.
While learning is an important mechanism of behavioral plasticity, behavioral plasticity also results from motivational and maturation processes. For example, females from a given butterfly population may respond to the same host plant cue in different ways (e.g., laying eggs or not), depending on the number of eggs that are currently matured. The more number of eggs that are matured, the more likely the female will lay eggs in response to the host plant cue. Thus, a response to an environmental cue (a behavior) is different as a result of reproductive state, rather than learning. It can be difficult to distinguish changes in behavior due to motivation or maturation from learning.
Learning involves but is not equivalent to information gain. Learning is more than just information gain. Animals acquire much, if not most, information without use of memory. Any behavioral response to a sensory stimulus is indicative of information gain. For example, a tap on our knee results in a knee jerk reflex. No learning is involved in this reflex, yet information about a mechanical stimulus impinging upon the knee has been acquired by our nervous system and responded to.
The changes in physiological state discussed above can also be used as source of information that can change behavior without requiring memory. For example, a bird's increasing hunger may indicate that food resources are locally scarce. If the bird chooses nest sites on the basis of relative food availability, its choice may involve learning but could also involve an estimate of availability derived from patterns in hunger level.
The equation of learning with information gain may reflect a human bias. For example, we say that we 'learn' what time it is or that we 'learn' that it is raining this morning. This language is not inappropriate, since humans store so-called declarative knowledge (or, 'knowledge of what' such as names and images of things or places) in a type of memory termed declarative memory. However, there is only limited evidence of declarative memory in nonhuman species. This form of memory is difficult to identify in nonhuman animals and its prevalence in animals is unknown.
Learning is not equivalent to synaptic plasticity. There is growing evidence that changes in the structural connections between neurons (termed synapses) mediate learning and memory in animals as different as vertebrates and insects. Nevertheless, an ecological and evolutionary perspective on learning need not and probably should not require a priori that a common mechanism underlie all examples of learning and memory. In fact, there is evidence of phenomena that resemble learning in every way, but which do not involve synaptic plasticity. For instance, feeding experience affects diet choice in locusts in a manner that resembles discrimination learning (a form of learning in which an animal learns to discriminate one food type from another). The effect of experience involves at least in part a taste-feedback mechanism in which the sensitivity of sensory receptors to nutrients in the hemolymph is adjusted. Conversely, synaptic plasticity simply means that synaptic function is not fixed with respect to inputs, and there are many kinds of synaptic plasticity unrelated to learning.
Was this article helpful?