Milner provokes and renews our quest for the integrative mechanisms and the autonomy of the mammalian brain, but, another thirty years on, we are still waiting for the empirical discoveries, and there remains a long way to go.
2. Milner's thesis does, however, distinguish itself by providing not merely a "reinterpretation of the educated salivations of a Russian dog," but in putting forward a model in which perception is strongly influenced by what an animal is planning to do. Indeed, this is Milner's closest approach to explicitly discussing any form of situated cognition or its importance in understanding ontogenetic, as opposed to phylogenetic, growth and development (Dickinson, 1997). One recurrent criticism of this volume, at least for me, was that although the natural history of any given species was regarded as critical for understanding its particular behavioral competences, there is no extension of this argument to include the importance of an individual's history of "task success" in explaining behavioral change throughout its own life-history.
3. Such a take-home message is possibly implicit in the discourse as the chapters continue, but given that the word "autonomous" is in the title, one might expect to find more of a guiding thread regarding the proposed move away from the reflexive, stimulus-response interpretations of the behaviorists. A further cause for concern (unfortunately shared with other good authors on this topic) is Milner's repeated use of black box models and accompanying nomenclature. The illustrations themselves are well presented (although typesetting and layout styles annoy frequently) and help focus orientation to the ideas laid out in the text -- itself an example of what the model describes! But the difficulties arise when one steps back to evaluate the reality of the system being modeled. On the one hand, Milner notes his predecessors' lack of neural substrate in explaining the mechanistic, neurological details of the operating nervous system underlying classical, instrumental and operant conditioning. On the other hand, we are here given input-output system components including "goal selectors," "response selectors," "response generators," "sensorimotor analyzers," etc., each component lacking in substrate (at least at this stage, see e.g., figs 2.2 - 2.3 and accompanying text). In fairness, however, Milner does refer the reader to later Chapters 8-9 for physiological accounts, but I am not sure they are really there either.
4. One of the reasons for this continuing mismatch is the nature of the interaction between the purely behavioral learning theorists and the cellular and molecular biologists interested in animal learning mechanisms. And although Milner attempts an integration of the knowledge acquired from all quarters, he obscures his quest by using the vocabulary of two of the most influential researches of his own life-history of intellectual growth and development (Lashley and Hebb). Whereas Hebb's valuable contribution of the idea of synaptic plasticity underlying learning behavior would be further developed with the advent of novel immunohistological techniques, molecular biology and cell membrane neurochemistry, Lashley's famous search for the engram continues to be put forward (as it is in this volume) and continues to fail to deliver (cf. Orbach 1999).
5. For example, Milner writes that, "contrary to the postulations of conventional learning theory... [in learning] associations take place that connect response plans (intentional activity) to engrams of stimuli involved in the performance, not the other way round." My argument here would not be with the context of the statement -- that motor apparati might guide selective attention, which in turn facilitates selective sensory sensitivity (this is fine with me) - but with the very use of "engram," a term which adds little to my understanding of the neural mechanisms underlying the behaviors seeking explanation here. Again, Milner may have done better in this instance to expand his thesis and discuss the adaptive significance of such engrams (be they "neural representations for selection," p. 23, or "a collection of sensory neurons that can be facilitated by an intention or a motivation," p. 120) rather than providing further promissory notes that to be discussed more fully later.
6. Insofar as passive recognizers of stimulus attributes are probably not instantiated in the nervous system, as Milner suggests, I am not convinced that his own characterization of more active ones are any more so. Although I will assuredly use chapters of this volume as a provocative senior honors teaching and tutorial text, I'm sure students will pick up on this point quite early in their discussions. As with the "motivation-driven attention" equated with the "activator of engrams" (see ch. 3), although the focus shifts from the stimulus (as object "out there") to the intent (to plan or do something), it does so only inasmuch as they are said to determine which associated engrams are aroused. The detailed structure of such engrams remains elusive, possibly tautological, and certainly begs for substrate instantiation at an earlier stage (perhaps even when they are introduced). That "engrams may become targets of attention" (p. 33), for example, following some successful performance or action, intention does not explain but merely redescribes the behavior post-hoc.
7. This apparent tautology (motivational intentions driving attention, informing motor action) at a variety of levels of description leads one to the "binding problem." As with Milner's short foray into the animal learning literature (in which he misses that the operant animal hardly needs to adapt to tasks of increasing levels of difficulty), a short discussion of the human clinical literature on the neural mechanisms of memory (important to short, intermediate and long-term binding issues) tends to concentrate on deficit behaviors. Milner, like most amnesia researchers, ignores the wealth of information available as a result of determining what amnesic patients can still do. There was the opportunity here to make more of the procedural versus declarative "knowledge/memory engrams" in that these patient's motor skills were found to be largely intact, together with the appropriate response activations and motor planning abilities. Such observations and their neurobehavioural sequelae might better inform the binding problem.
8. As with most of the redescriptions put forward since William James's (1899) observation about our difficulty in coming to understand perceptual constancy in the world of "blooming, buzzing confusion" detected by our sensory systems, Milner offers no new light other than to reiterate this problem: what does seem likely is that the activities in different areas of the brain representing various features of a stimulus all acquire associations with ongoing activity that identifies the episode when the stimulus is present. Hints are given about the problems inherent in neural time codings (not the least of them in determining the synchrony of sound and light sources coming from the same object) and in spatial configuration (both in external objects and in the coordination of the body parts with which one tries to make avoid or contact). The highly nontrivial example of the neurobiology of hand-eye coordination and the real link between the developing sound/motor coordination system (babies listening to themselves vocalize) bring home the important point that we learn the sensory consequences of our actions. But are these the core issues that the enigmatic engrams are intended to explain?
9. In these early years of the 21st Century, we know that all animals have evolved some ways of deriving benefit from the effect of their experiences. Activity-induced changes are known to alter the effectiveness of existing nerve synapses and their extended connectivities throughout the life-history of the animal; but the true autoregulatory mechanism(s) underlying this ability remain elusive for now. Reading this work of Milner's, I was frequently reminded of Pribram's (1971) "Languages of the Brain," in which he admitted that his prediction concerning the search for the engram succeeding where Lashley had failed had itself failed. Milner provokes and renews our quest for the integrative mechanisms and the autonomy of the mammalian brain, but, another thirty years on, we are still waiting for the empirical discoveries, and there remains a long way to go.
Dickinson, A.R. (1997) Hierarchical Organisation in Serial Search Tasks by Cebus apella monkeys. Unpublished PhD Thesis. University of Edinburgh.
McGonigle, B. O. (1991). Incremental intelligent systems by design. In J. Arcady-Meyer and S. Wilson (Eds) From Animals to animats. Cambridge: MIT Press.
McGonigle, B. O. & Chalmers, M. (1996) The Ontology of Order. In L. Smith (Ed.), Critical readings on Piaget. London: Routledge.
Milner, P.M. (1999a) The Autonomous Brain. Erlbaum, Mahwah NJ
Milner, P.M. (1999b) Precis of "The Autonomous Brain" PSYCOLOQUY 10(71) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.071.autonomous-brain.1.milner http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.071
Hendriks-Jansen, H. (1995) Catching Ourselves in the Act: situated activity, interactive emergence, evolution and human thought. MIT Press. Cambridge, Mass.
James W. (1890) The Principles of Psychology. Macmillan, London, UK.
Orbach, J. (1999) Precis of: The Neuropsychological Theories of Lashley and Hebb. PSYCOLOQUY 10(029). ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.029.lashley-hebb.1.orbach http://www.cogsci.soton.ac.uk/psyc-bin/newpsy?10.029
Pribram, K. (1971) Languages of the Brain. Prentice-Kall, Englewood Cliffs, New Jersey.