In behaving animals, the process or the outcome of learning are measured by monitoring changes in performance. But learning and performance are not equivalent. Learned information may remain latent until the appropriate conditions emerge for its expression as a change in performance ('latent learning'; Tolman 1932). Earlier in the twentieth century, orthodox "behaviourists attempted to shy away from this common knowledge. They claimed that only overt behavioural acts (definition 3) are legitimate psychological data, whereas inferred implicit changes in the potential to behave are not. The distinction between performance and competence, however, is now taken for granted in memory research. It does complicate the life of the experimenter who studies learning and memory in the behaving organism. For how could one be sure whether an apparent failure on a memory test is due to faulty learning, feeble memory, or impaired performance under the test conditions? The impaired performance could be due to trivial causes, such as defective sensorimotor capabilities, but also to more elusive causes, such as lack of "retrieval cues, inappropriate "context, diverted "attention, subthreshold motivation, too little or too much arousal,1 latent alterations in the ability of "stimuli to control behaviour, and more. The distinction among these types of causes requires smart test designs, which take into account factors that could hinder the expression of behaviour (for selected examples of pitfalls and how to circumvent them, see "experimental extinction, "state dependent learning, "transfer). In other cases, the brain itself may simply need more time to surrender its new knowledge ("insight).
The distinction between performance and competence should be a source of concern in dealing with "reduced preparations as well (definition 1). These preparations range from isolated ganglia ("Aplysia) to brain slices and neuronal cultures ("long-term potentiation). Cellular analogues of learning, such as a lasting "synaptic facilitation, may remain dormant under certain conditions yet become apparent under others. This may due to inappropriate test conditions such as nonpermissive "ionic composition of the solution in which the brain slice is immersed, unfavourable neuromodulator^ state ("neurotransmitter), or impaired cellular metabolism ("nutrients). Similarly, even if we knew the representational code in a given neuronal circuit, and were in principle able to infer learning by noting the alteration in the "internal representation in that circuit, we might still fail to identify the change because of certain nonpermissive states of the system or the context. The differentiation between competence and performance may hence pop up at various "levels of analysis, from the behaving organism to its individual synapses.
In some cases competence is not aptly transformed into performance simply because the system can do well without exploiting its full "capacity. This is nicely demonstrated in language. We know many more words than we use in daily life. This is why educated adults tend to estimate their vocabulary at a figure that is only 1-10% of the actual value (Seashore and Eckerson 1940). Whereas the vocabulary used in routine daily activities ranges from a few hundreds to a few thousands words, depending on education and profession, and goes up to 8000-20 000 in literary works,2 the number of distinct words in printed school English (excluding derivatives and compounds) is about 89000, of which an average 6-year-old child commands already no less than 13 000 (Pinker 1994), and a high school graduate 27000-53 000 (Nagy and Anderson 1984). This implies that tests to quantify skill should not only be permissive for the expression of this skill, but also properly designed to allow the expression of its capacity. Training and testing conditions that allow for the expression of maximal performance may, however, yield 'ceiling effects'. Testing under ceiling conditions is not appropriate for measuring delicate alterations in performance, such as caused by learning or "development, because it could mask the effect of the experimental treatment.
Performance has multiple roles in the different "phases in the life history of a memory. It is a key element in "instrumental conditioning, acquired via stimulus-response (S-R) contingencies.3 Repetitive performance is essential for training on "habits and skills; this, among others, underlies the use of simulators in training (Hammerton 1967; "transfer). And in solving complex problems, performance itself may actually be an essential step in the "algorithm: the subject (an organism or a computer) tries the problem, attempts a solution, which produces a new problemsolving strategy, which is then used to tackle the problem again, and so on. The overall approach involves a sequence of transformations from one attempted strategy to another, each emerging from the one that just preceded it. This is called 'learning by doing' (Anazi and
Simon 1979), a sort of on-the-job training. Performance is also, of course, the embodiment of retrieval. The distinction from competence notwithstanding, performance is the ultimate measure of learning both in the field and in the laboratory (Richardson-Klavehn and Bjork 1988; Martin and Bateson 1993). However, as performance involves activation of neural circuits and therefore "plasticity on the one hand, and adaptive interaction with a dynamic outside world on the other, performance in retrieval may also induce modification of the trace upon its use (re-"consolidation). Thus usually, performance Pj of a task is not an exact replica of performance Pj_p because the performance itself actuates learning. Virtuoso musicians should surely attest to that.
Much has been learned in recent years about "internal representations that underlie motor performance and learning, and about distinct brain regions that monitor the behavioural performance and its deviation from the desired output (Carter et al. 1998; Kawato 1999; "dopamine, "planning). Prevalent theories of motor control and motor learning propose that the brain generates, stores, refines with experience, and executes internal models of the world, that mimic the input/output characteristics of the specific motor act, and uses them on-line to calculate the desired motor commands (Jeannerod 1994; Kawato 1999). Imagery may engage such internal models without culminating in overt performance (Jeannerod and Decety 1995). Furthermore, there are reports that imagery can substitute for real action in motor training; under such conditions of 'learning by imaging', the distinction between overt and covert performance becomes even fainter (e.g. Yaguez et al. 1998).
At least in certain types of mental operations, performance is also believed to be instrumental in understanding the world. There is evidence that in some "perceptual and "recognition tasks, the brain construes the perceived behavioural act by activating internal representations that are capable of performing that same act. This possibility has been suggested specifically by the so-called 'motor theories' of vocal recognition. These theories, which apply to human speech, mammalian calls, and "birdsong, propose that speech sounds are perceived and distinguished by tacit knowledge of the vocal gestures used in their production (Liberman et al. 1967; Peterson and Jusczyk 1984; Williams and Nottebohm 1985). If this is the case, then the take-home message is that those who cannot perform cannot understand (for a similar conclusion see also "observational learning). This conclusion, however, should not be taken too orthodoxly and seriously; we can surely enjoy a melody even if we sing notoriously out of tune.
Selected associations: Habit, Insight, Mnemonics, Observational learning, Skill
1The observation that appropriate performance requires an optimal level of arousal is called 'the Yerkes-Dodson Law' (Yerkes and Dodson 1908).As all the other 'laws' of behaviour, this is not a law but rather a pragmatic 'generalization. The Yerkes-Dodson law should be kept in mind when an attempt is made to improve memory, either by behavioural methods ('mnemonics) or by drugs ('nootropics). Hence excessive training could result in a weaker memory ('spaced training), and taking stimulants before an exam may damage attention and performance. This was clearly realized by Maimonides (1180): 'the righteous way is the median measure'.The Yerkes-Dodson law is hence a special case of the 'Maimonides Law', itself a reformulation of the old wisdom that preaches for taking the golden path in life (aurea mediocritasin Latin).
2Shakespeare used 15000 words, Milton 8000, but Italian Grand Opera enchanted audiences with 800 words only (Seashore and Eckerson 1940).
3But see Deese (1951) and Solomon and Turner (1962) for selected cases in which overt performance is not essential to obtain an instrumental response.
Was this article helpful?