Abstract
From a personal point of view, an unexpected consequence of the various COVID lockdowns was a certain existential angst not previously experienced. Obsession with serial position curves and millisecond differences became replaced with worries about the nature of science, mind-body problems, and neurally plausible accounts of thinking. Against this backdrop, the opportunity to review The cooperative neuron presented itself, and this was grasped with both hands. There is absolutely no implication here, though, that The cooperative neuron came about merely as a by-product of government-imposed restrictions: it clearly reflects a lifetime’s engagement with academic research and university teaching. Much ground is covered; ranging from hyperpolarization-activated cyclic nucleotide-gated (HCN) channels to, eventually, the meaning of life (well Life’s Purposes Section 10.3) and, to be honest, in places the terrain is hard-going.
Where better to start than with McCulloch and Pitts (1943/1990, henceforth M-P) and the thought that individual neurons must be doing logic. According to Anderson (1995), the original M-P neuron can only be in one of two states (ON or OFF), has a fixed threshold, receives input from excitatory synapses all of which have the same weight, and sums such activation linearly. At first glance, such a formalism sits comfortably with the notion that neurons “integrate and fire.” Strictly speaking though, this is not right because although an M-P neuron can receive activity from an inhibitory synapse, any such input closes the neuron down. Nonetheless, it is commonplace to assume that integrate and fire neurons that signal the Truth/Falsity of a basic logical predicate are key to understanding “how brains can be minds” (front inside dust cover). According to The cooperative neuron however, this view is just plain wrong.
The contrasting picture that emerges in The cooperative neuron comes about from taking a hard look at the nature of pyramidal cells: the most common neuron in the cortex. We need to replace the constraints of thinking in terms of cold logic and “integrate and fire” with the notion of cooperative, context-sensitive processing in which pyramidal cells interchange information about that which is currently relevant. If only we can be more accepting of the nature and importance of these kinds of cells, then we have a panacea for the whole of psychology and beyond.
The first four chapters provide the necessary scene setting and groundwork that cements the foundations for what follows; the foundations for what Phillips calls cellular psychology. In contrast to neuropsychopharmacology, the aim is to relate “basic hypothetical constructs . . . to their cellular foundations” and to take such constructs and their cellular foundations and relate these to phenomenology (p. 20). The early material is brisk, in places repetitive, and aimed at the uninitiated. I found it to be both informative and interesting. However, I am no fan of footnotes and each chapter has them, and these are collected at the end of the chapter rather than in page footers. In the early chapters, there is also quite a bit of forward referencing—in later chapters backwards referencing also comes into play. Collectively, this brought about quite a lot of page flipping. On several occasions, I found myself following the signposts, becoming absorbed in the corresponding material and then struggling to back up and pick up the thread.
A perhaps minor point, but references are made to coloured figures that are in fact printed in monochrome (at least in my copy of the book). I found this mildly irritating because I felt that to do the material justice, I needed to concentrate hard. Not being able to see the “short red lines” in Figure 3.1 (p. 46), the blue dots in Figure 3.6 nor the red and blue areas in Figure 3.7 only added to the workload. Things were much easier when the coloured figures are actually coloured. I should also admit to not knowing where the thalamus is, and although Figure 2.1 is provided, I could only just about make out “Pulvinar of Thalmus”(?). Figure 3.1 is key. A URL is provided to the original figure and that helped (it is coloured) but nonetheless I was unable to decipher some of the additional symbols and numbering used in the printed figure. Things to bear in mind for a second edition, perhaps?
The description of pyramidal cells is intimately woven with a description of the lamina nature of the cortex. There are three types of pyramidal cells that are defined, roughly, in terms of which layer of the cortex they inhabit, where they project to, and where they receive inputs from (p. 37). However, what is key is that pyramidal cells are context-sensitive two-point neurons (Section 3.3). They are distinctive in having two separate sets of inputs; respectively, those from apical dendrites (far from the cell’s soma) and those from basal dendrites (near to the cell’s soma). The argument is that the different inputs serve different purposes. From what I could gather, “Basal dendrites of pyramidal neurons in perceptual regions receive inputs that usually convey information about current sensory stimulation,” for example “the presence of a face or even a particular face” (p. 49). But this left me wondering what pyramidal neurons outside of perceptual regions do. Things are less murky regarding what apical dendrites do because almost the rest of Chapter 3 is devoted to what functions apical dendrites serve. What is striking is how these modulate the cell’s activity. Input to the apical dendrites “amplify or attenuate the transmission of information extracted from the other inputs” (i.e., the basal dendrites) but only when coincident excitation in both sets of inputs. The upshot is that we have cooperative context-sensitive neurons that “can cooperate by using one subset of inputs to amplify or attenuate the transmission of information extracted from another subset of their input.” (p. 68).
Given this re-conceptualization of what neurons are actually doing, Phillips then goes on to use this to address (among other things), “Various State of Mind and Brain” in Chapter 4, “Mental Life” in Chapter 5, “Language and Thought” and several developmental issues in Chapter 6, various pathologies in Chapter 7, and information processing in the brain in Chapter 8.
Before confronting Chapter 5, I became waylaid by Section 4.6 Mental State and Conceptions of Consciousness. Five common assumptions concerning consciousness are set out (p. 101) and Phillips takes issue with all five. I became absorbed with some of what is written and became intrigued by three of the assumptions. Assumption 1 is that “all experiences are ‘of’ something other than themselves.” Assumption 3 is that “we have an adequate notion of self . . . that has the experience.” Assumption 4 is that “all direct experiences must be conscious or unconscious.”
With respect to Assumption 1, although this is presented as being a common assumption, without further analysis, it is not at all clear what is being assumed. It seems to me that there is a sensible distinction to be drawn between having a sensation of warmth and consequently realising that the car seat heater is on. We might want to then try to decide whether the perception is of the warmth or the car heater. In contrast, Phillips merely wants to defend the position that some experiences (e.g., seeing colour, feeling pain) may come about in the absence of a relation to an external entity. He also asserts that, “some thoughts are created in the act of thinking them” (p. 101). It is only fair to confess that I now struggle to pass on what the actual point of substance is being made in this regard—the brain creates experiences?
In addressing the concept of the “self,” Phillips discusses the notion of an enduring entity “a self” that has experiences and he treats this with great scepticism. I was expecting some discussion of Hume’s view that, “the self is nothing over and above a constantly varying bundle of experiences.” 1 However, Phillips merely concludes (p. 316) that, “I see no evidence for a single enduring ‘self’ in the context that guides processing and learning at a cellular level.” I am not sure who would wish to argue otherwise, although, on reflection, this might be a veiled rebuttal of supervised learning systems. But things do get a little weird when we are told “there are grounds for supposing that there may be a primitive form of ‘consciousness’ in the brain stem, and that this may include some form of ‘pre-reflexive self-awareness’” (p. 130). What the actual grounds are and what pre-reflexive self-awareness is remains to me, at least, something of a mystery.
Assumption 4 bears repetition: “That all direct experiences must be either conscious or unconscious.” There are many vexed issues here. For instance, I thought the use of “direct” here was unnecessary and begged questions regarding what indirect experiences might be. Generally, though, we need to proceed with great care. To be told that, ‘Subjective experiences are system properties” (p. 98) is not in itself that helpful and even though we are reassured that “No neuron can be a mind.” (p. 98) in the closing passages (p. 319) it turns out that there is a “form of cooperation in which cells know about and cooperate with others . . .” (italics added). All of this brought back with some force just how challenging it is to try and convey anything sensible about the nature of experience in terms of language (see for instance, McGinn, 1997, Chapt. 5).
My more substantive point is that my intuitions lead me to conclude that experiences are by definition conscious. Of course, there are those who have it otherwise (see for instance Searle and his discussion of unconscious pain, 1994, pp. 164–167), but Phillips centres his discussion on differences between being awake, dreaming and being in “dreamless slow wave sleep.” A case is built that these are inexorably linked with particular type-distinct states of apical dendrites of pyramidal cells (see Table 4.1). In being new to this material, I did find the neuroscience regarding these different mental states to be very interesting regardless of my concerns about the philosophy.
Chapter 5 is almost a monograph in itself: Eighty-nine pages long supplemented by nine pages of 113 footnotes. The chapter begins with discussion of various aspects of both unimodal visual perception and cross-modal visual/auditory perception but quickly returns to the topic of consciousness. Now further evidence is used to bolster the case that, “the neural bases of conscious human experiences are a broad class of macroscopic thalamocortical events in which the outputs of a large, dynamically selected subset of cooperative neurons are amplified or generated by inputs to their apical dendrites!” (p. 107). Having thus nailed consciousness, attention and working memory are turned to next. Phillips states that “selective attention can operate via cooperative context-sensitivity by directly activating apical dendrites as well as by disinhibiting them.” (p. 141). In discussing visuospatial short-term memory, we are told that this, “involves persistent poststimulus activity of cells in frontoparietal regions . . . (and this) “amplifies sensory response to input from one or a few selected locations in space. It does so in much the same way that as attention to selected aspects of current simulation, which also involves apical function.” (p. 164).
The material has been quoted at length because it conveys the kind of analysis that is adopted throughout. Various amounts of cellular evidence is used to bolster such claims, but as a cognitive psychologist, I was both intrigued and frustrated. In discussing attention, passing reference is made to the work of Spratling and colleagues (p. 142; e.g., Spratling & Johnson, 2004), and it seems clear that the ideas are chased down in terms of computational modelling couched at the level of pyramidal cells. However, because I was not prepared to consult the primary references, I was left with no clear sense of what had been achieved or how. The lack of a proper treatment of the models left this reader having to take Phillips at his word. I have similar gripes with the treatment of Heeger and Mckey (2019, pp. 171–173), and in particular, the Principle of Free Energy Reduction Section 9.2.1. To understand fully these topics, the primary sources need to be consulted.
Part of the argument conveyed in Chapter 5 is that both attentional and short-term memorial effects are due to apical function. Of course, the details are key, but being reasonably familiar with some of the recent literature on directing attention in visuospatial short-term memory (see e.g., Atkinson et al., 2018), this kind of speculation was of no help in my attempts at understanding the behavioural data. There are effects of memory and there are effects of attention, and these dissociate in interesting ways. I need to be convinced that a more detailed examination of apical function can provide the necessary explanatory apparatus for why.
It would be remiss not to note that Phillips devotes a whole chapter (Chapter 9) to Difficulties and Unresolved Issues. The fact that none of these occurred to me as I digested the prior material suggested to me that I am not at all well-placed to assess the neuroscience. What I do note however is that Phillips concedes that, “Not all psychological phenomena described as involving context imply cooperative context-sensitivity as defined here, but many do.” (p. 189). It is unfortunate that this not followed up by an in-depth exploration of the differences between those that do and those that do not.
As is probably clear, I am conflicted in appraising this work. It is without doubt that this book is extraordinarily ambitious in its sweep; an immense amount of detail is conveyed, and very broad proclamations are made. I am sure that those wishing to learn about basic neural processes that relate to mental life will gain a lot from engaging with the text. The cooperative neuron may also act as a manifesto for those dissatisfied with traditional ideas about neurocomputation, but I simply don’t know. I came away with some feelings of unease, however. I think Phillips builds an interesting case for the presence of cooperative context-sensitive neurons. What I was less comfortable with were the examples of linking these to higher-order properties. I remain cautious about the general framework for thinking about the “cellular foundations of mental life” as developed in the book. We can characterise human society as reflecting context-sensitive cooperation, and I am open to the possibility that this arises from (because of?) context-sensitive cooperation at the cellular level. May be, may be not?
