The change in performance during training that is taken to represent the progression of learning

Memories are like people—they are born, live, and die. Acquisition is their moment of birth. The other major phases in the life history of a memory are "consolidation (if it is ever to become a long-term memory), storage, "retrieval, and extinction ("experimental extinction, "forgetting). Depending on the context of discussion, 'acquisition' implies a temporal phase (definition 1, e.g. Stillings et al. 1987); or a process that takes place during this phase (definition 2, e.g. Tulving 1983); or a change in "performance that reflects this process (definition 3, e.g. "behaviourism). This change in performance is quantified by an 'acquisition curve' or 'learning curve', in which performance is plotted against the amount of practice (e.g. Skinner 1938; e.g. Figure 41, p. 144). Commonly, the "subject is said to have completed the acquisition of the task if its performance has reached a preset "criterion, such as time to reach the goal in a "maze or a certain probability of success on a discrimination problem (e.g. "delay task). The process of acquisition was termed 'engraphy' by Semon (1904), meaning the engraving of an "engram, but 'engraphy' has never caught on. 'Acquisition' is sometimes used as a synonym for '"learning', but the latter term has a broader meaning and usage.

Acquisition is composed of subprocesses. The first is 'encoding', which in general refers the conversion of a message from one language, or code, to another. 'Encoding' is frequently used in the learning literature as a synonym for 'acquisition', but this is unsatisfactory, because there is more to 'acquisition' than 'encoding'. In neuronal encoding, information is transformed into the neuronal codes used in computation and representation (Churchland and Sejnowski 1992). This information arrives from either the external or the internal world. In the first case, the electromagnetic, mechanical, or chemical information is converted via the sense organs into neuronal activity. In the second case, information from the body itself is conveyed by specialized neuronal circuits, or via body fluids in the form of chemical messages (hormones) that evoke neuronal activity. No information can be handled by the central nervous system without first being encoded into the appropriate neuronal code. Encoding is thus involved in brain activities that do not necessarily culminate in the acquisition of a memory, such as on-line processing of information ("attention, "percept), or control of ongoing physiological routines. For a memory to be born, an additional process, of initial 'registration' ('recording'), is also needed. This permits the "internal representations of transient "stimuli, once formed, to become or induce an engram. From what we know from physiology and psychophysics, the decay time of transient representations is in the subsecond range (Dudai 1997b, see also 'encoding time' in Ganz 1975; "cell assembly, "percept, "phase). The registration mechanisms hence differentiate transitory from lasting internal representations, where 'lasting' is anything that is significantly longer than the aforementioned decay time.

How much time does acquisition require? This depends on the learning "paradigm and protocol. It is convenient to distinguish 'instant' from incremental ('repetitive', 'rote') acquisition. Instant acquisition refers to single-trial learning. This takes place in certain situations of intense aversive conditioning ("conditioned taste aversion, "fear conditioning); in some types of "imprinting; in the formation of "flashbulb memories; and probably in some other situations, in which acquisition curves have a step-function shape (e.g. "insight). In contrast, incremental acquisition refers to situations in which information accumulates over multiple experiences to construct the memory (Pavlov 1927; Skinner 1938; Hebb 1949; Dudai 1989). Gradual acquisition of "habits and "skills is such a case. The repetitive practice is expected to involve gradual modification of internal representations over hours, days, even months. But does incremental acquisition involve accumulative modifications that are restricted to the original representation formed at the beginning of training? This assumption might be naive. Internal representations are expected to form dynamic distributed networks ("cell assembly). Therefore, a more realistic view is that recurrent discrete events of acquisition and consolidation, that stem from each accumulative experience, alter existing internal representations that encode the information in question, but at the same time generate new representations and link them to the old ones ("palimpsest).

Ample data, supported by learning theory, indicate that whatever happens in acquisition, in terms of perceptual "cues and cognitive processes, determines not only the lifespan of the resulting memory, whether short or long (Craik and Lockhart 1972; Baddeley 1997), but also how efficiently will this memory be "retrieved in due time. Two influential concepts that reflect this notion will be mentioned here. One is the 'encoding-specificity principle' (Tulving 1983). It states that memory performance is best when the cues present at retrieval match those present in acquisition. The other is termed '"transfer-appropriate processing' (Morris et al. 1977). It states that memory performance is best when the cognitive processes invoked at retrieval (say, semantic as opposed to phonetic processing in verbal tasks) match those used in acquisition.

Multiple approaches are used to investigate the neurobiology of acquisition. Cellular physiology, neuropharmacology, neurochemistry, and molecular biology are all applied to dissect the molecular and cellular mechanisms involved. Candidate 'cellular acquisition devices' are "ion channels and membrane "receptors on synaptic terminals that receive the teaching input, itself encoded in ion currents and "neurotransmitters ("Aplysia, "long-term potentiation). A substantial amount of information is also available on the processes downstream from the synaptic membrane, that involve activation of "intracellular signal transduc-tion cascades, and couple acquisition to consolidation. We even seem to start to understand in molecular terms why is it that in many learning situations, distributed training with intercalated intervals between repetitive acquisition trials, is more efficient than massed, continuous training, in which acquisition mechanisms are expected to function nonstop ("spaced training).

Brain areas and neuronal circuits that subserve acquisition have been identified in "habituation, "sensitization, "classical, and "instrumental conditioning in a variety of "simple or less-simple "systems (e.g. "Aplysia, "classical conditioning, "conditioned taste aversion, "Drosophila, "fear conditioning, "honeybee). In recent years, "functional neuroimaging has made a remarkable contribution to the identification of brain systems that subserve acquisition in the human brain (e.g. Nyberg et al. 1996; Fletcher et al. 1997; Tulving and Markowitsch 1997; Buckner and Koutstaal 1998; Epstein et al. 1999; Fernández et al. 1999). The circuits that acquire information about a memory vary with the type of memory, but a few general conclusions emerge from the studies so far: (a) acquisition of "declarative memories engages widely distributed areas, which include modality specific "cortex, and in addition supramodal areas, particularly in the mediotemporal lobe ("hippocampus, "limbic system); (b) these areas partially overlap brain areas that later retrieve the learned information; and (c) in some studies it was possible to show a correlation between the activation of an identified brain region during the training experience and the subsequent ability to remember this experience. For example, the ability to remember verbal information could be predicted by the magnitude of activation in the left prefrontal and temporal cortex during the training (Wagner et al. 1998b). It is not yet known, however, which of the activated areas is indispensable for acquisition ("criterion), which area is causally related to the strength of the engram, and what are the specific roles of each of the areas in the encoding and registration of information in the first milliseconds and seconds after engraphy has been triggered.

Selected associations: Consolidation, Experimental extinction, Retrieval, Transfer

0 0

Post a comment