Peter M. Milner (2001) Stimulus Equivalence, Attention and the Self: a Punless Reply. Psycoloquy: 12(016) Autonomous Brain (9)

Volume: 12 (next, prev) Issue: 016 (next, prev) Article: 9 (next prev first) Alternate versions: ASCII Summary
PSYCOLOQUY (ISSN 1055-0143) is sponsored by the American Psychological Association (APA).
Psycoloquy 12(016): Stimulus Equivalence, Attention and the Self: a Punless Reply

Reply to Gellatly on Autonomous Brain

Peter M. Milner
McGill University
Department of Psychology
1205 Dr. Penfield Ave.
Montreal, QC. CANADA
H3A 1B1


I was flattered to discover that my book stirred the reviewer's imagination, but his approval leaves me with little excuse for rejoinder. I have seized the opportunity to clarify my views on stimulus equivalence and the role of the response system in controlling perception. At Gellatly's request I have enlarged on my treatment of self.


association of ideas, attention, behaviour model, intention, motivation, self, serial order
1. So far, The Autonomous Brain does not appear to have stirred up much in the way of controversy among reviewers. Nobody has got hot under the collar either at my attack on Hebb's empiricist reverberating cell assembly or at my materialist and nativist biases. I think this bodes well for psychology in the new century but it makes for less exciting reading in a talk-back journal like this one. Gellatly's (2000) informative review offers me few grounds for argument but I will do what I can.

2. Gellatly lists the main features of the book and generally approves of them. Nevertheless I was somewhat taken aback by a comment in his first paragraph in which he states that I think psychologists show an excessive interest in visual perception because of an obsession with stimulus equivalence. As a matter of fact I think the main reason for our interest in vision is that we are primates, and vision is our most important sense. The message I intended to convey is that it is unwise to try to work out a theory of perception based on the visual system because stimulus equivalence is sure to crop up, and it proved a hard nut to crack. This was disingenuous on my part. It would have been more candid to admit that for pedagogical reasons I thought it wiser to explain a simpler perceptual system, such as olfaction, in which individual receptors do a good deal of the work of stimulus identification. The more abstruse problem of stimulus equivalence could then be explained separately.

3. Although Gellatly's interpretation is not the one I had intended, my surprise was at how close it came to the reality of my own experience. I have never considered the stimulus equivalence problem to be either unimportant or uninteresting. In fact it was my stubborn determination to solve it that distracted me from my interest in motivation a quarter of a century ago and started me on a course that led to The Autonomous Brain. I spent most of a sabbatical year exploring spatial frequency analyzers, perceptron networks and other blind alleys until it slowly dawned on me that Hubel and Wiesel (1959) had solved the problem in their lab ten years earlier, and no-one had noticed. So I wrote a paper (Milner, 1974) to tell everyone, but still nobody paid much attention. Some people are still trying to solve the stimulus equivalence problem. In the course of writing the paper I ran into a further problem. Once one has found a way to recognize a shape independently of its size and position in the visual field, how does one know where it is? That led to consideration of reciprocal sensory pathways, attention, and what is now called the binding problem. Speculation about olfactory perception could never have been so rewarding.

4. One of the oldest and strongest ideas in behavior theory is that representations of stimuli acquire connections with response control neurons. I believe that associations in the opposite direction, between response control neurons and stimulus representations are equally, perhaps even more, important. Thus it is important to make the direction of the association as clear as possible. A sentence like the one near the end of Gellatly's paragraph 3 "Especially important are associations between assemblies representing stimuli or ideas and those representing response plans or intentions." is ambiguous and will be interpreted as associations in the stimulus - response direction by the vast majority of readers. This would be less serious if the next sentence did not require the alternative interpretation. Selective attention involves control of the sensory system by an intention.

5. The idea that intention determines attention has been around for more than a century but the fact that it implies association between response neurons and the attentional paths of the sensory system seems never to have fired the popular imagination. Because a hungry rat will make a perilous leap towards a card displaying a triangle, it is commonly assumed that the image of a triangle has acquired control of the animal's jumping response. The rat actually jumps because it is hungry and, after a certain amount of trial and error, it believes there is food behind the card. It looks for a triangle to discover in which direction to jump. Recognition that the concept of an object is often present before the object is perceived solves many of the problems of perception. It means that the central end of the sensory path is already established, eliminating promiscuous and irrelevant associations that could occur with activity initiated only peripherally.

6. In his paragraph 4 Gellatly expresses disappointment that I do not develop the idea of the self more fully and prompts me with a few questions. I should perhaps preface my response by pointing out that my early training was in engineering and the closest I have been to a philosophy course was an introductory course in psychology. I suppose I introduced the idea of a "self" as a way of dramatizing the inherited mechanisms most animals, and even some robots, possess in order to prevent overlapping of goals and merging of responses. The idea also serves to illustrate the difference between dualist and monist models. A robot that transmits its visual input to a controller who interprets it, thinks about it, remembers it and then delivers instructions to the robot's effectors, is a robot for dualists. A monist's robot is one that has goals and the required circuitry to achieve them built in (e.g. deploying its solar panels when batteries run low and seeking shade when overheated).

7. I wrote that subjectively decisions are made by the "self". Human beings are aware of urges that require a conscious choice to be made - tea or coffee; cigarette or gum; get a job or go to college. Only people verbalize these options and they are probably the only beings that speculate about the nature of the entity that chooses. In that sense I believe only human beings are fully self-aware. Self-awareness implies an ability to be conscious of one's own mental processes.

8. I took a fly as an example of an organism that can make decisions but is unlikely to have a concept of "self". Near the other end of the scale, chimpanzees that have been taught to communicate have ways of referring to themselves. The chimpanzee is almost the only animal that recognizes that it sees itself in a mirror; most animals treat the reflection as another animal (Gallup, 1968). But it is impossible to say whether a chimpanzee attributes the self to anything other than its body. I am predisposed to believe that rats and mice are conscious of what goes on around them, but who can know what, if anything, passes through their consciousness when they hesitate at a maze choice point, presumably trying to remember which is the way to food? Are they aware that they are testing the motivational associations of different response plans? I rather doubt it. In fact I doubt whether this search for the self, though it may be instructive and even mildly entertaining, will ever reach its goal. We can, however, look for the decision making circuits with some hope of eventual success. Perhaps we will be able to demonstrate that the decision mechanism is a neural structure while the "self", when present, is a neural activity, possibly in a different structure.

9. I doubt if these random remarks answer Gellatly's questions about the self to his entire satisfaction. If he requires fuller answers to them from better informed sources I suggest that he should pose them at greater length in a target article. It is a topic that I am sure will elicit very lively comment


Gallup, G.G. Jr. (1968) Mirror image stimulation. Psychol. Bull., 70, 782-793.

Gellatly, A. (2000). Smell assemblies and the brain. PSYCOLOQUY 11(101) psyc.00.11.101.autonomous-brain.8.gellatly

Milner, P.M. (1974) A model for visual shape recognition, Psychol. Rev., 81, 521-535.

Milner, P.M. (1999a) The Autonomous Brain. Erlbaum, Mahwah NJ

Milner, P.M. (1999b) Precis of "The Autonomous Brain" PSYCOLOQUY 10(71) psyc.99.10.071.autonomous-brain.1.milner

Volume: 12 (next, prev) Issue: 016 (next, prev) Article: 9 (next prev first) Alternate versions: ASCII Summary