Abrupt improvement in the performance on a task

Some types of learning, such as the acquisition of "skill, progress gradually, through numerous repetitions. This is termed 'incremental learning', or 'rote learning' (Hebb 1949). In contrast, other types of learning, both in "real-life and in laboratory setting, occur abruptly, following a step function. Two examples are "flashbulb memory and "conditioned taste aversion. There is, however, a type of abrupt learning that differs from o these examples. It cannot be described as fast acquisition of information about an on-line event. Rather, the "subject is presented with a problem, which it finds difficult to solve. After a typical period of either fruitless overt attempts or behavioural silence, suddenly, a solution comes to mind. We all are familiar with these situations. There is even a special term that conveys the subjective flavour: the 'AHA! experience' (Kaplan and Simon 1990; Sternberg and Davidson 1995). When we watch animals under situations that seem to involve sudden realization of a solution to a problem, we tend to "anthropomorphize and conclude that they also have their 'insight', uttering 'AHA!' in doglish or chimpanzeesh.

Some cases of alleged insight are anecdotal. It is told that the chemist Kekule suddenly saw in reverie the structure of the benzene ring, although doubts were raised whether the story is true (Gruber 1995). The most famous of all 'AHA!' experiences is that of Archimedes, who allegedly jumped naked from his bath shouting 'Eureka' (Greek for 'I have found it'), after suddenly realizing a "method to determine the amount of alloy mixed with the gold in the crown of the king of Syracuse. Again, the story is surely refreshing, but already Galileo considered it implausible (ibid.). As science cannot rely on anecdotes, various protocols were developed to demonstrate insight under controlled conditions. The following are two "classic examples. Kohler (1917, 1925) reported a number of 'insightful' or 'intelligent' experience-dependent behaviours in the chimpanzee. In a typical experiment, he placed a chimpanzee behind an array of vertical bars, and a heavy stone with a cord tied around it on the other side (Figure 37). Food was attached to the cord halfway between the anchoring stone and the bars. The cord was inserted through the bars in an angle that prevented the chimp from reaching the food. The only way to get the reward was to shift the cord between the bars until it formed a straight angle with the array of bars. The chimp tried at first to pull the cord, but in vein. Suddenly she somehow realized that she can take the cord in one hand, pass it around the bar to the other hand, and hence move it from one inter-bar spacing to another, step by step, till it formed the required angle, culminating in happy end. Individual chimps differed in the kinetics of their response, but most reached the right solution.

Other classic 'insight' experiments were carried out by Maier (1931) on humans. A typical one is 'the pendulum experiment'. Subjects were introduced into a room that had two strings hanging from the ceiling, and a number of other objects, including poles, pliers, and

oooooooooooooo

Food

Fig. 37 A *classic insight experiment. Kohler (1917) placed a chimpanzee behind an array of vertical bars, and a heavy stone with a cord tied around it on the other side. Fruit was attached to the cord halfway between the stone and the bars, and the cord was inserted through the bars in an angle that prevented the *subject from reaching the reward (a). The only way to get it was to shift the cord between the bars until it formed a straight angle with the array of bars. Just pulling the cord was of no avail. Suddenly the chimp realized that she can take the cord in one hand, pass it around the bar to the other hand, and hence move it, step by step, from one inter-bar spacing to the other, till it formed the required angle (b). (Adapted from Kohler 1917.)

Fig. 37 A *classic insight experiment. Kohler (1917) placed a chimpanzee behind an array of vertical bars, and a heavy stone with a cord tied around it on the other side. Fruit was attached to the cord halfway between the stone and the bars, and the cord was inserted through the bars in an angle that prevented the *subject from reaching the reward (a). The only way to get it was to shift the cord between the bars until it formed a straight angle with the array of bars. Just pulling the cord was of no avail. Suddenly the chimp realized that she can take the cord in one hand, pass it around the bar to the other hand, and hence move it, step by step, from one inter-bar spacing to the other, till it formed the required angle (b). (Adapted from Kohler 1917.)

cords. The subjects were asked to tie together the two hanging strings. However, the distance between the strings was too large. After a while, the subjects suddenly realized that they can tie the pliers to one of the strings, swing it like a pendulum, and catch it in its up-swing while holding the other string. Reports of 'insight' are not confined to anthropoids. Over the years, cases of apparent 'insight' were described, although some also disputed, in cats in problem boxes and rodents in "mazes (e.g. Walker 1983). An interesting example of 'insight' in lower vertebrates involves pigeons who pushed a box in order to be able to climb and peck a hanging banana (Epstein et al. 1984). In all these experiments, it is important to discern between chance successes occurring via repetitive attempts of "instrumental learning, and a sudden solution to a problem after behavioural silence.

Reports of 'insight' made "behaviourists sad and Gestaltists happy.1 The mere idea that learning involves internal reorganization in the brain in the absence of explicit "stimulus-response coupling contradicted the orthodox behaviouristic tenet that only public behaviour is the subject matter of psychology and that internal processes should be ignored. Attempts have been made either to play reports of'insight' down, or to explain 'insight' by rather complex chaining of stimulus-response contingencies (Keller and Schoenfeld 1950; Bower and Hilgard 1981). In contrast, the Gestalt trusted that the nature of "perceptual and mental parts is determined by the whole, and that enquiry into the mind should consider global organization and proceed top-down (Kohler 1925; Koffka 1935; Hochberg 1998). The idea that mental structures are restructured to achieve a new meaning in the "context of previous knowledge was exactly in line with what they were preaching for. With time, 'insight' became a focus of interest in cognitive psychology in relation to information processing and problem solving (e.g. Weisberg and Alba 1981; Kaplan and Simon 1990; Ohlsson 1994a,fc; Sternberg and Davidson 1995). Several "models have been proposed for insightful behaviours. They involve elements such as reshuffling and recategorization of building blocks of prior problem-related knowledge, the sudden identification of "cues, the use of multiple heuristic solutions and the identification of invariants in such solutions. Among these elements, the need for prior problem-related knowledge stands out in species far away on the phylo-genetic scale (e.g. Epstein et al. 1984); again, it appears that in order to learn something, we must already know a lot ("a priori). However, as noted by Kaplan and Simon (1990), knowledge is a two-edged sword, as inflexible knowledge may guide the search for the solution astray.

A major question concerning insight is whether it a special type of learning, differing in its computational strategy, the "algorithms and their biological implementation from rote learning. For example, is insight restricted to higher processing "levels because it seems to affect 'global' cognitive structures? And are there any specific circuit properties that play a part in insight only? Hebb (1949), for example, who considered insight as the most advanced form of adult learning, did not think that its basic mechanisms differ from those of rote learning. So far, one of the difficulties in comparing insight to incremental learning was the traditional use of different types of behavioural paradigms—problem solving in the study of insight, skill acquisition in the study of incremental learning. The use of certain "perceptual learning tasks, involving visual detection and discrimination, could provide a solution, because these tasks tap into elements of both skill learning and insight (Ahissar and Hochstein 1997; Rubin et al. 1997). In these tasks, if the task is made easy, learning "generalizes over "dimensions of the stimuli, matching the properties of high-level visual areas, whereas when the task is made difficult, learning shows little "transfer if at all, typical of low-level perceptual skill. Under certain conditions, performance on these skill-like tasks shows sudden improvement, resembling insight, either on the particular task, or on the ability to solve the type of tasks ("learning set). It was suggested that this abrupt improvement on 'low-level' learning depends on increments in 'high-level' knowledge (Rubin et al. 1997). Although this conclusion could prove to be "paradigm-specific, it does cast doubts on the generality of type-distinction between rote learning and insight.

The attempt to separate abrupt from gradual learning brings to mind the so-called 'Sorites paradox' (soros, Greek for 'heap'; Williamson 1994). It is attributed to Eubulides of Miletus, a contemporary of Aristotle, and it goes like this: does one grain of wheat make a heap? Do two grains? Three? Ten thousands? Where is the transition point? For those remote from agriculture, the 'bald man' version may evoke more empathy: Is a man with one hair on his head bald? With two? Three? If the addition of any single hair is not critical, one is led to admit that a man with 10 000 hairs is bald. The boundaries of the terms 'heap', 'bald', 'incremental', or 'abrupt learning' are therefore vague. Attempts to distinguish insight from incremental learning resemble attempts to consider '"flashbulb memory' as inherently different from other episodic memories. Naturally, we are impressed by the extreme cases. But in real life, a spectrum should be expected of time-scales, interactive levels of processing, and complexity of "internal representations. Insight may involve latent increments in knowledge, and rote learning may involve 'microin-sights'. Sharp "taxonomies may exist only in the eye of our cognition.

Selected associations: Acquisition, Binding, Delay task, Learning

1For more on the Gestalt school of psychology see *binding.

Positive Thinking As The Key To Success

Positive Thinking As The Key To Success

Download this Guide and Discover How To Find And Monetize on Your Expertise And Strengths. Inside this special report, you'll discover: How positive thinking is one of the key factors in a successful life. Five ways and tools to help you stay positive. Use these to help you keep on track. Case studies that'll inspire you to stick to your dreams. Plus much, much more.

Get My Free Ebook


Post a comment