Ruth Garrett Millikan: White Queen Psychology and Other Essays for Alice.

Cambridge, MA: MIT Press, 1993.

Anthony Chemero

Indiana University, Bloomington
[1] It is an unfortunate fact about philosophers that they are very rarely convinced to change their minds about the positions they hold, even when faced with arguments that ought to convince them to do so. Since I am as guilty of this as anyone else, I was surprised to find that an essay included in Ruth Millikan's new collection convinced me of something that I had previously thought impossible -- that a correspondence theory of truth can be developed within a strictly naturalist framework (that is, without reference to unanalyzed propositions, states of affairs, or the like). The essay that did so is 1990's "Truth rules, hoverflies and the Kripke-Wittenstein paradox," and it would probably convince more people if it were not just as difficult to understand as Millikan's first book, the notoriously impenetrable Language, Thought and Other Biological Categories (1984). It is this difficulty, along with the importance of Millikan's work, that makes White Queen Psychology such a welcome collection. The essays included here, with a few exceptions, are considerably more readable than her first book, but have lost little in their depth of insight.

[2] The fourteen papers, five of them new, defend and expand upon the biologically based theories of semantics and cognition laid out in Language, Thought and Other Biological Categories. The most important points in the previously published papers concern Millikan's views about functions, representation, and truth. A thing's function, she claims, is that which it was designed to do, what it would do if it performed properly in the environment in which it evolved (in what Millikan calls "normal conditions"). When choosing from among the infinity of possible functions to ascribe to something, choose the one the performance of which enhanced the survival value of the organism throughout its evolutionary history. So, to cite a famous example, the function of the magnetosome found in certain bacteria is not to orient the bacteria toward magnetic north or toward the sea bottom (both of which it actually does), but rather away from the oxygen-rich surface water that is toxic to the organism. Also, a thing need not on average perform its function (how many sperm actually fertilize an egg?).

[3] A representation, Millikan claims, is a thing that can only perform its function when it corresponds to some state of the world. To understand representations correctly we must consider three things: the representation producer, the representation consumer, and the representation itself. The content of a representation depends upon what the representation consuming part of the organism takes the representation to represent. To determine the content of a representation we must focus on the conditions under which the consumer uses the representation successfully; that is, under normal conditions. The function of a representation producer is to produce representations that correspond (according to a certain rule) to aspects of the normal environment. The representation producer will fail most of the time in historically non-normal circumstances. So the fact that frogs will eat BBs in a laboratory setting does not mean that small projectiles cause the frog's representation producer to make a representation of "fly or BB here now". The content of that representation is, instead, "fly here now," but because it was produced in non-normal conditions (by BBs, in the laboratory) it is a false representation. As for the representation itself, Millikan holds something like Wittgenstein's picture theory from the Tractatus. At a suitably abstract level, the representation must picture what it indicates. They must admit of significant transfomations which accord with the variations of the thing that is represented. That is, representations must be what Millikan calls "articulate," (everyone else calls this property "compositional"). For if they were not compositional, they couldn't be about states of the world and so could not be true or false. Hence the frog representation is of "fly here now" and not simply "fly."

[4] In order to argue for a correspondence theory of truth, Millikan claims that our language abilities are not a collection of dispositions, but are rather a set of biological mechanisms whose function it is to cause us to conform to rules. If viewed as such, grasping correspondence truth rules is no more problematic than grasping verificationist ones. She differentiates between following proximal rules and following distal rules. A proximal rule is one the following of which depends only upon incoming sensory information and an intact physiology. Following a distal rule depends, in addition, upon the actual state of the world. In many cases, the function of proximal rules is to effect the following of distal rules, while the function of distal rules is to aid the organism in surviving in its normal environment.

[5] Consider again the bacteria that has magnetosomes. The magnetosome's function is to cause the bacteria to follow the following proximal rule: 'If there is a source of a magnetic field nearby, move toward it.' The function of the proximal rule is to effect the following of this distal rule: 'If the surrounding water is rich in oxygen, move toward water that has less oxygen' (oxygen is poisonous to these bacteria). Note that following the proximal rule will often fail to lead to following the distal rule (in the presence of a clever experimenter with an electromagnet, for example). Verificationist truth rules, such as those favored by anti-realists like Dummett and Putnam, must be assertibility rules. These rules, Millikan claims, are proximal rules; they are based solely upon information available to the senses. Correspondence truth rules are distal rules because they are based upon conditions in the world and not just incoming sensory information. Imagine that truth rules are distal correspondence rules. Then they must be backed by proximal assertibility rules, whose function it is to cause conformity with the distal truth rules.

[6] As we noted above, following the proximal rules often does not lead to conformity to the distal rules because distal rules refer to factors outside the individual. So, conforming to proximal assertibility rules will often fail to lead to conformity to distal correspondence truth rules. This accounts for the normativity of the notion of truth. The conditions in which biological systems function always make reference to external factors; that is, those conditions that have been present throughout evolutionary history when the object fulfilled its biologically proper function. So, proximal assertibility rules which make no reference to external conditions could not define semantics, they must be the more or less reliable means used to follow correspondence truth rules. So, we need not share proximal assertibility rules to share truth rules (how many ways are there to make a map?). We do, however, seem to share distal correspondence truth rules. Millikan concludes from the above that if we can grasp assertibility rules, we can also grasp correspondence rules.

[7] Three of the new essays are concerned with individualism in the philosophy of psychology. Two of these, "What is behavior?" and "The green grass growing all around," constitute a whole called "A philosophical essay on ethology and individualism in psychology." In these she argues that behavior is the functional form of an organism's actions, so it can't be described adequately without reference to the environment and speculation about biological functions. It follows from this that biological facts don't straightforwardly supervene on physical facts. This is because behavior is defined only by referring to biological functions, and since a thing's functions are determined historically, identically structured physical objects can have different functions. So the behaviors of physically identical objects (say me and my molecule-for-molecule twin created just now by God) might need very different explanations. Furthermore, because of the importance of the normal environment in understanding behavior, there really is no principled distinction between an organism and the normal aspects of its environment. As Millikan puts it, "I no more carry my complete cognitive systems around with me as I walk from place to place than I carry the U.S. currency system about with me when I walk around with a dime in my pocket" (170). The mistake made by psychological individualists, then, is that they focus solely on the insides of people's heads when a large portion of a human cognitive system is outside in the world.

[8] Millikan describes the title essay, an 84-page monograph called "White Queen Psychology; or, The Last Myth of the Given," as a diatribe against "meaning rationalism," the basis for individualism. Meaning rationalism, she claims, consists of the claims that identity and difference of meaning, univocity, and meaningfulness are all epistemically given to the thinker. Millikan describes three varieties of meaning rationalism, differentiated according to how sameness of propositions is determined, and unleashes a series of externalist arguments against each of them. The conclusion she draws from the falsity of meaning rationalism is that meanings are not epistemically given, so there is no problem with, for example, believing contradictory propositions. Because of this, logical possibility cannot be known a priori. Rationality, like meaning, ain't in the head; it is, instead, a property of a properly integrated "head-world system."

[9] Millikan draws some conclusions about intentional rational (=cognitive) psychology based upon this fairly radical anti-individualism. Psychology must become a more self-consciously ecological science. The mechanisms that humans use to form beliefs and desires are biological mechanisms, and so have biological functions. Since biological functions are determined only with reference to evolutionary history and the normal environment, it makes no sense to try to understand human psychology as a series of inputs and outputs in laboratory conditions. "[P]redictions of the motions of individual heaps of human cells under random conditions is obviously not what human psychology is about. It is about the (wider) proper development and operation of human behavioral systems under conditions normal for carrying out their biological functions" (170). Further, since biological mechanisms are not guaranteed to fulfill their functions (remember the sperm cell), even intact humans in favorable environmental conditions may acquire many false, redundant and contradictory beliefs. So intentional psychology should not expect to find laws of behavior. It should focus on, and try to explain, those cases where human cognitive systems function properly in their normal environment, just as the physiologist's job is to explain properly functioning hearts and kidneys inside the body.

[10] Millikan is, I think, absolutely correct in her assessment of the future of human psychology. Futhermore she is correct in claiming that "especially in the case of those behaviors that are intentional actions, reference to mechanisms of inner representation will play a large role in [behavioral] explanation" (169). Problems will arise, however, if one tries to explain any behavior by referring to the type of abstract, picture-like representations Millikan discusses. Pictures are static things, even if one can perform transformations upon them; there is nothing static about the world humans and other animals live in. So the type of representations employed in explanations of an organism's ability to get around in its environment cannot be simple copies of incoming sensory information (whether current or recalled from memory). Representing the world seems to be an active process -- one capable of creating vast amounts of new information, ignoring the greater part of incoming sensory information and performing extremely rapid changes of perceptual state. No mere picture or set of pictures can do this. In having a relatively static theory of representations (whether picture-like or sentence-like), Millikan is part of a long-standing and virtually unquestioned tradition. This tradition needs to be reconsidered, and Millikan's suggestion that we treat psychology as a truly ecological science is a good basis upon which to do so.


Millikan, Ruth (1984). The Language of Thought and Other Biological Categories. Cambridge, MA: MIT Press.
Anthony Chemero
Indiana University, Bloomington

The Electronic Journal of Analytic Philosophy 2:6 (1994)
ISSN 1071-5800
Copyright 1994