The generation modification or selection by the brain of internal representations of future actions and their anticipated consequences

'It is a poor sort of memory that only works backwards', said the White Queen to Alice (Carroll 1865). But mocking at the memory of the visitor to Wonderland was rather unjustified. For even without pushing modern physics to its limits, we do have a kind of 'memory of the future' (Fuster 1992). This is planning.1 This capability should not be belittled: in spite of its alleged reputation, even Destiny, speaking through the Delphic oracle, only rarely dared to give clear predictions of the future (Fontenrose 1978).

'Planning' originally referred to the design of ground plans for the purpose of planting, itself a word derived from planta, Latin for 'sole of the foot'. Definition 1 is the generic one; Definition 2 assumes volition.2 This is what people commonly have in mind in considering planning. The distinction between definitions 1 and 2 touches philosophical issues that exceed the scope of this brief discussion; suffice it to note that definition 2 applies to living organisms (myths and deities notwithstanding), yet in the future will fit smart robots as well. Definition 3 rephrases definition 1 while focusing on the "reduced concept of internal representations, which is central to discussion of "learning and "memory in this book.

Planning could be manifested in a gamut of markedly different behaviours, ranging from fast motor actions to long-term strategic plots. It is therefore methodologically useful to classify it ("taxonomy), using "criteria such as the behavioural domain, complexity, timetable of the anticipated action, flexibility, and the trigger for the plan.

1. Behavioural domain. An elementary dichotomy distinguishes motor planning, referring to motor acts and their anticipated consequences, from cognitive planning, referring to cognitive narratives and their mental, behavioural, and social consequences. Multiple taxonomies are possible within the behavioural domain "dimension. Related dimensions are the complexity of the planned goal, and the explicitness of planning ("declarative memory).

2. Time to attain the goal. From what has been said above it becomes apparent that different planning acts could refer to very different temporal spans, ranging from seconds on the one hand to years on the other. Compare, for example, a tennis player planning a tricky serve vs. a college student planning a career. In considering the shortest time spans, two issues are noteworthy. One is the automaticity of the plan. This will be further discussed in 3 below. The other is the transformation from the present to the future. As the cognitive present in not infinitesimally small, plans of brief motor acts raise the question what is 'present' and what is 'future'. Based on psychophysical and physiological data, 'present time' appears be in the order of ~ 100 ms (this estimate is task dependent; e.g. Thorpe et al. 1996; Helenius et al. 1998; "percept). In motor tasks such as orienting and reaching, neuronal activity is detected in the brain several hundreds of milliseconds prior to the actual execution of the act (Georgopoulos et al. 1989). This could therefore be construed as a candidate neuronal correlate of bona fide elementary planning (Georgopoulos 1994; Andersen et al. 1997; Flanagan and Wing 1997). Within this narrow temporal window, there is ample time to modify the plan by intervening with the activity of the circuit during the planning phase (Groh et al. 1997; "method).

3. Innateness and flexibility. But how much of a shortlived motor plan is really a plan, rather than the deterministic unfolding of an automatic response? Even the most sophisticated, imaginative cognitive plans are expected to depend to some degree on "a priori constraints on our perceptual and cognitive faculties, which lead to automaticity of certain elements in the behaviour. Yet clearly, some plans are more constrained by innate predispositions than others. It is phylogeneti-cally advantageous to limit the anticipated alternative outcomes of fast motor plans that are essential for survival, by relying on innately encoded internal representations that can be selected within fraction of a second. The most elementary brief motor plans could be said to differ from elementary reflexive responses in that the former include a component of active selection and decision making by the "system (definition 2), whereas the latter are solely deterministic reaction. The reflexive response could still be fine-tuned by experience, but the action pattern is a given once triggered. This distinction between 'reflexive' and 'nonreflexive' is convenient yet shaky; what appears as a 'voluntary' planned act might still be an intricate reflex shaped by complex input (Luria 1962).

4. Trigger. Plans ranging from very simple to rather complex ones (e.g. ambushing a prey) could be triggered by a single 'release "stimulus' (Lorenz 1981). In contrast, complex cognitive plans may be initiated by "retrieved or endogenously generated representations.

In spite of their immense heterogeneity, in all mental plans the internal representations are expected to include a representation of a hierarchical set of sequential actions and of the anticipated outcome (Miller et al. 1960). These organized sets of actions, termed 'schemata' (Bartlett 1932), draw from past experience—be it encoded in innate or learned responses or both. In other words, although plans refer to the future, they always draw from either the species' or the individual's past.

Naturally, the investigation of planning in laboratory setting is limited to tasks that are completed within a few minutes or a few hours at most. Systems that are particularly useful for the cellular and system analysis of elementary planning in the "monkey brain involve limb, head, or eye movement (e.g. Georgopoulos 1994; Thach 1996; Andersen et al. 1997; Flanagan and Wing 1997; Zhang and Barash 2000). These studies focus, depending on the specific task, on the role of motor and posterior parietal "cerebral cortex, as well as the "cerebellum. Relevant to this area of research is also the investigation of the role of neuromodulatory systems ("neurotransmitter, "dopamine) in the selection among alternative response patterns.

In the study of planning in humans, the tests commonly involve puzzles that could be solved by mentally testing sequences of moves ahead of time, such as the Tower of London task (Shallice 1982; Figure 56). In this task, three beads, one red, one green, and one blue, have to be moved in a minimal number of steps from the initial configuration on three sticks of unequal length to the designated goal position. The problems vary in complexity, from two to five moves. More recent versions of the task involve the manipulation of coloured shapes on a computer screen (e.g. Dagher et al. 1999). Other types of tasks that tap into human planning involve simulation of "real-life situations, such as financial planning (Goel et al. 1997). All in all, the combined data from neuropsychological and "functional neuroimaging studies in patients with circumscribed brain lesions and in normal volunteers, implicate frontal cortical areas in the multiple aspects of cognitive planning (Luria 1962; Shallice 1982, 1988; Fuster 1995a; Goel et al. 1997; Owen 1997; Bechara et al. 1998; Dagher et al. 1999; Figure 56; for an example of "modelling of planning in cortex, see Dehaene and Changeux 1997). This cortex is 'where the past and future meet' (Fuster 1995) to integrate past experiences, both declarative and nondeclarative (Bechara et al. 1997), in planning ahead.

Selected associations: A Priori, Algorithm, Dimension, Prospective memory, Working memory

1A related type of memory of the future is 'prospective memory, which is discussed separately.

2Volition refers in this context to autonomous decision making by a 'system regardless of its state of 'conscious awareness. Hence 'planning' according to definition 2 could either be 'declarative or nondeclarative.

Fig. 56 The Tower of London task is used to test planning because it is possible to solve it by mentally testing sequences of moves ahead of time, before actually executing them. (a) In the original version (Shallice 1982), three beads, one red, one green, and one blue, have to be moved from the initial configuration on three sticks of unequal length to the designated goal position in a minimal number of steps. The problems vary in their complexity, from two moves to five moves. More recent versions of the task involve the manipulation of coloured shapes on the computer screen rather than beads and sticks (e.g. Dagher et al. 1999). (b) The *performance on the task of *control, healthy volunteers and of brain lesioned patients, plotted as the percentage correct on the first attempt to solve the problem vs. the complexity of the problem. LA, left anterior brain lesion; LP, left posterior lesion; RA, right anterior lesion; RP, right posterior lesion. The patients with left anterior brain lesions, involving the left frontal lobe, were dramatically impaired. (Adapted from Shallice 1982).

0 0

Post a comment