Learning

Although learning can be understood as a change in an organism's capacities or behavior brought about by experience, this rough definition encompasses many cases usually not considered examples of learning (e.g., an increase in muscular strength brought about through exercise). More important, it fails to reflect the many forms of learning, which may be distinguished, for example, according to what is learned, may be governed by different principles, and may involve different processes.

One form of learning is associative learning, in which the learner is exposed to pairs of events or stimuli and has the opportunity to learn these pairings -- which event or stimulus goes with which (see CONDITIONING). The study of associative learning has led to the discovery of numerous learning principles applicable to species as diverse as flatworms and humans, and behaviors as different as salivation and the onset of fear. Investigators have also learned a great deal about the biological bases for this form of learning (see CONDITIONING AND THE BRAIN).

Another crucial form of learning involves the acquisition of knowledge about the spatial layout of the organism's surrounding -- these COGNITIVE MAPS can include the locations of food sources or of dangers, the boundaries of one's territory, and so on. The acquisition of this spatial knowledge often involves latent learning: The organism derives some knowledge from its experiences, but with no immediately visible change in the organism's behavior. (The latent learning does become visible later on, however, when the organism finds occasion to use what it has earlier learned.) In many species, this learning about spatial layout can be extraordinarily sophisticated, allowing the organism to navigate across great distances, or to remember the locations of hundreds of food caches, or to navigate by dead reckoning, with little or no reliance on sensory landmarks (see ANIMAL NAVIGATION and ANIMAL NAVIGATION, NEURAL NETWORKS).

Spatial learning may be considered a special case within a broader category of learning, in which an organism gains ("memorizes") information about its environment (see MEMORY). In humans, this information may be derived directly from firsthand experience, or indirectly, from what one reads or hears from others. This information may then be used later on for some memory-based report (e.g., a response to the question, "What happened yesterday?"), or as a basis for modifying future action. In any of these cases, one encodes the information into memory during the initial exposure, and then retrieves this stored information later on. The initial encoding may be intentional (if one is seeking to memorize the target information) or incidental (if the learning is a by-product of one's ordinary commerce with the world, with no intention of learning). Similarly, the subsequent use of the information may involve explicit memory (if one wittingly and deliberately seeks to use the stored information later on) or implicit memory (if the stored information has an unwitting and automatic influence on one's subsequent behavior; see IMPLICIT VS. EXPLICIT MEMORY).

Still another form of learning is skill learning, in which one learns how to perform some action or procedure, often without any ability to describe the acquired skill (also see MOTOR LEARNING). In this case, one is said to have acquired "procedural knowledge" (knowing how to carry out some procedure), as opposed to "declarative knowledge" (knowing that some proposition is correct). It should be emphasized, however, that skill learning is not limited to the acquisition of motor skills (such as learning how to serve a tennis ball or how to ride a bicycle). In addition, much of our mental activity can be understood in terms of skill acquisition -- we acquire skills for reading, solving problems within a particular domain, recognizing particular patterns, and so on. Thus, for example, chess masters have acquired the skill of recognizing specific configurations of chess pieces, a skill that helps them both in remembering the arrangement of the game pieces and (probably more important) allows them to think about the game in terms of strategy-defined, goal-oriented patterns of pieces, rather than needing to focus on individual pieces (see EXPERTISE and PROBLEM SOLVING).

Skill learning can also lead to AUTOMATICITY for the particular skill or procedure. Once automatized, a skill can be run off as a single, integrated action, even though the skill was initially composed of numerous constituent actions. The skilled tennis player, for example, need not focus on wrist position, the arch of the back, and the position of the shoulders, but instead launches the single (complex) behavior, "backhand swing." This automatization promotes fluency among the constituents of a complex behavior, dramatically decreases the extent to which one must attend to the various elements of the behavior, and thus frees ATTENTION for other tasks. On the other hand, automatic behaviors are often inflexible and difficult to control, leading some to speak of them as "mental reflexes."

A further form of learning is INDUCTION, in which the learner is exposed to a series of stimuli or events and has the opportunity to discover a general rule or pattern that summarizes these experiences. In some cases, induction is produced by the simple forgetting of an episode's details and the consequent blurring together in memory of that episode with other similar episodes. This blurring together is, for example, the source of our knowledge of, say, what a kitchen is likely to contain. Investigators refer to knowledge acquired in this fashion as "generic" or "schematic knowledge" (see EPISODIC VS. SEMANTIC MEMORY and SCHEMATA).

In other cases, induction results from a more deliberate judgment process in which one actively seeks to generalize from one's previous experiences, a process that seems to rely on a relatively small number of strategies or JUDGMENT HEURISTICS. For example, subjects in many studies seem to rely on the assumption that the categories they encounter are relatively homogeneous, and this encourages them to extrapolate freely from the sample of observations made so far, even if that sample is relatively small, and even (in some cases) if warnings are in place that the sample is not representative of the larger category (see CATEGORIZATION).

Some aspects of induction seem to be governed by highly specialized domain-specific skills. One clear example is provided by LANGUAGE ACQUISITION in the small child. The human infant appears to be well prepared to induce the regularities of language, so that language acquisition is relatively swift and successfully achieved by virtually all children, independent (within certain boundary conditions) of the child's individual abilities or circumstances. The same learning skills, however, seem irrelevant to the acquisition of information in other domains (also see COGNITIVE DEVELOPMENT and DOMAIN SPECIFICITY).

Finally, let us note still other forms of learning: Many species are capable of learning through IMITATION, in which an action is first observed and then copied. A number of species display imprinting, in which a young organism learns to recognize its parents or its conspecifics. Human learning often also involves DEDUCTIVE REASONING, in which one is able to discover (or generate) new knowledge, based on beliefs one already holds. In some cases of deduction, one's reasoning is guided by relatively abstract rules or principles. In others, one's reasoning is guided by a specific remembered experience; one then draws an analogy, based on that experience, and the analogy indicates how one should act, or what one should conclude, for the current problem (see CASE-BASED REASONING AND ANALOGY).

Thus the term learning plainly covers a diversity of phenomena. But having now emphasized this diversity, we should ask what these many forms of learning have in common. At a general level, some principles may apply across domains -- for example, the importance of acknowledging task-specific learning skills, or the possibility of latent learning, not immediately manifest in behavioral change. At a much finer-grained level, it is likely that similar processes in the nervous system provide the substrate for diverse forms of learning, including, for example, the process of LONG-TERM POTENTIATION, in which the pattern of interaction among neurons is modified through experience. Similarly, it is plausible that connectionist models may provide powerful accounts of many of these forms of learning (see COGNITIVE MODELING, CONNECTIONIST). In between these extremes, however, we may be unable to formulate general "laws of learning," applicable to all learning types.

Learning

See also

Further Readings