Jackendoff on Consciousness

Andrew Brook


In `How Language Helps Us Think', Jackendoff explores some of the relationships among language, consciousness, and thought, with a foray into attention and focus. In this paper, we will concentrate on his treatment of consciousness. We will examine three aspects of it: 1. the method he uses to arrive at his views; 2. the extent to which he offers us a theory of consciousness adequate to assess his views; and 3. some of the things that we might need to add to what he offers to achieve an adequate theory.

1. Introduction

`How Language Helps Us Think' offers some valuable and novel reflections on the relationship of language to thinking. Since it turns out, on Jackendoff's view, that many of the ways in which language helps us think is by helping us to gain conscious access to our thoughts, he is soon led back to his old interest in consciousness. Jackendoff is no stranger to research in consciousness; his well-known 1987 book (hereafter C&CM) was built around what he calls the intermediate-level theory of consciousness. The intermediate-level theory is the view that what we are conscious of is material at an intermediate level of processing: for language and material expressed in language, the level of phonetic representation, which is between the level of acoustic input and conceptual content; for vision, the level of Marr's 2 1/2D visual representation, which is, roughly, things purely as they appear, after the two dimensional retinal array has been given a third dimension but before the application of general knowledge to produce a full 3D representation. Put another, what we are conscious of is intermediate between sensory input and conceptual structure and content. Of course, `intermediate' just expresses an image; the relationship could just as well be visualized as on top, the metaphor Freud and many others favoured, at the centre, or underneath. What matters is the idea that what we are conscious of is what occurs at an intermediate level of processing.

The paper under consideration both fleshes key aspects of that model out and goes beyond it in important ways. In connection with the latter, Jackendoff now thinks that being able to bring the contents of a thought to consciousness makes a positive contribution to the soundness and efficiency of thinking about it; in C&CM he thought rather the opposite, that "having language doesn't enhance thought, it only enhances the experience of thought" (p. 14, current paper).

Another is the substitution in the current work of a notion of valuation for what he used to call affects. He has in mind things like deciding whether a representation is externally or internally generated, of a real or imaginary object, of something familiar or novel. One is a bit taken aback to find him saying that valuations "have not to my knowledge been singled out for attention elsewhere in the literature" (p. 19). To a philosopher's ear, many of the things that Jackendoff calls valuations look like nothing more than a subclass of what Russell called propositional attitudes, the taking up of an attitude to the contents of a perception or a thought or an imagining or whatever. However, not all valuations are propositional attitudes and Jackendoff may be after something else. He certainly knows the philosophical literature on propositional attitudes.

Perhaps the following explains his comment, though I am not sure. By `propositional attitude', Russell had in mind mainly the epistemic attitudes: doubt, belief, questioning, etc. By valuations, however, Jackendoff has in mind things like attitudes about the source, reality, familiarity, or affect to be taken up to a representation or even the difficulty of processing it cognitively. Some of these things are propositional attitudes, the ones involving the taking up of an attitude to something that has propositional form, but some of them are not -- and it would be true to say that the latter have not received much attention.

In these remarks, I will confine myself to Jackendoff's view of consciousness. In his view, consciousness is central to his view of the relationship between language and thinking, his thinking about it has changed in interesting ways, and he is not always as clear about the phenomenon as he might be. I will urge that he needs to say more about the phenomenon than he has done so far. To argue this, I will look at three things: the method he uses; what he offers by way of a theory of consciousness, which will reveal an apparent confusion; and some reasons for thinking that an adequate theory of consciousness must go farther than Jackendoff has gone so far.

2. What is Jackendoff's method?

In connection with consciousness, Jackendoff tells us that his central question will be: Which brain phenomena appear in consciousness and which do not? (p. 4) Let's start with a sample of the claims that he makes in the course of answering this question:

1. "The forms of thought [entailments and other semantic relations] are not ... available to consciousness. ... More generally, I am inclined to think that thought is never conscious." (p. 7)

2. "We become aware of thought taking place ... only when it manifests itself in linguistic form... The thoughts expressed by ... words [like] kill and die are too general, too abstract to be conveyed by a visual image." (p. 8)

3. "Only by having a linguistic modality is it possible to experience the steps of any sort of abstract thought. For example, one can't directly see that monkey Z is a member of X's kin-group. ... Similarly, the notion of retaliation ... is unavailable to nonlinguistic consciousness." (p. 13)

4. "My sense is that consciousness has nothing at all to do with processing difficulty or executive control of processing: that's the function of attention." (p. 15)

5. "We can pay attention only to something that we are conscious of. ... Having linguistic expressions in consciousness allows us to pay attention to them." (p. 16/17)

6. "Language is the modality of consciousness in which the abstract and relational elements of thought are available as separate units." (p. 18)

(1) - (6) are a fair sample of Jackendoff's claims about consciousness. Are they correct? And by what method does he set about to persuade us of their truth? (1)-(3) and (6) flow directly from the intermediate-level theory of consciousness, so any reason to doubt them will also be a reason to doubt the intermediate-level theory in general.

Concerning (1), Jackendoff seems to think that it is just obvious that, for example, rhyming relationships are "transparent to awareness" and meaning relationships are not. He does have an argument to back the intuition: it has taken us a long time to find the mechanical principles underlying sound thought, he says, whereas until recently nobody was even interested in the principles underlying rhyming. But this argument is quite weak: that we do not know the principles of something does not imply that we are not directly aware of the thing itself. Until very recently nobody had any idea of the principles underlying wood, either; but that did not stop us from being aware of tables and other wooden objects. Given the weakness of the argument he gives, I think Jackendoff's main reason for accepting (1) is that he thinks that it is obvious. To me it is far from obvious: when I `see' that if X is a bachelor, X is an unmarried man, who is to say that the relationship between the two notions is any less `transparent to consciousness' than the relationship between `slew' and `moo'? (I have deliberately chosen words that sound alike but do not look alike.)

(2) is connected to three interrelated ideas: that the thoughts expressed by words like kill and die can enter consciousness only if expressed in a medium; that such thoughts are too general, too abstract to be conveyed by visual images or any similar medium; so the medium must be linguistic. On the first idea: what is the argument that we cannot be conscious of a thought directly, that it must be expressed in some medium? The same claim is at the heart of (3), that "only by having a linguistic modality is it possible to experience the steps of any sort of abstract thought". Recall Jackendoff's claim in connection with his example there: "one can't directly see that monkey Z is a member of X's kin-group". I don't see that he offers any reason to accept either claim -- except an appeal to the kind of intuitions that we discussed in connection with (1).

The second idea has any plausibility only if we conceptualize visual images on the model of sense-data, Lockean ideas. In this regard, Jackendoff seems to think that he has Wittgenstein (1953) on his side, that the latter also held that visual images could not express thoughts. This is far from clear. Wittgenstein certainly attacked a certain conception of visual images, namely the 18th-century sense-data conception. And he certainly held that appearing in an image could not be what is crucial to something being a thought or the thought it is, what is distinctive to thoughts. But these ideas are very different from any claim that a thought could not be expressed in a visual image at all, a visual image of any kind. Where does Wittgenstein urge anything like that? On the question itself, are we so sure that a notion such as killing could not be expressed in a visual image of the kind created by, say, a Picasso?

The issue of whether thought can be expressed in any non-linguistic medium relates to (6), the claim that language is the modality of consciousness in which the abstract and relational elements of thought are available as separate units (p. 18). One wonders how this view would fare in the face of a painting such as Guernica. The images and other elements of this painting seem to express "abstract and relational elements" of some thoughts, indeed rather important thoughts. And they clearly exist as separate units. I imagine that compositionality is behind Jackendoff's claims here. Compositionality is indeed one of the features of language that contributes most to its expressive power, but some nonlinguistic media of thought/communication can surely manifest at least low level compositionality.

On the third idea connected to (2): is it clear that all thoughts of which we are aware appear to us in linguistic form? There seem to be some ready counter-examples, the phenomenon of having a thought on `the tip of one's tongue' but not being able to find the words to express it being an obvious one. It may be urged that all we are aware of here is, say, the thought of a thought, and that we can put that into words. Tempting or no, this cannot be right; we are often in the position of knowing what the thought we cannot capture in words relates to, what other thoughts it excludes -- in short, we are aware of a lot of the semantic information contained in it. We just aren't aware of it in words. (Curiously enough, in C&CM, Jackendoff takes tip of the tongue situations to support his view (pp. 290-1). He thinks that when we cannot find the words, we are also unaware of the thought -- of anything, indeed, except the bare fact that we have a thought. This just seems to me wrong.)

That leaves (4) and (5). They both concern attention. In (4), we are told that it is the function of attention, not consciousness, to deal with processing difficulty and exercise "executive control" over processing, and in (5) that "we can pay attention only to something that we are conscious of", which in turn requires a "linguistic expression" of the item (p. 16/17). There seem to be counter-examples to both claims.

Against (4), consider consciously focusing on something so intensely that one is hardly aware of anything else for a time; think of trying to solve a mathematics or logic problem. Here our consciousness of the problem certainly seems to play a role in the difficult processing of information that goes on in connection with it. Against (5), there is the case of high-level expertise in a complex activity, for example playing the piano. It seems plausible to suggest that when a pianist plays something difficult, she is giving attention to all the elements in the music. Yet she is not conscious of many of them, not individually at any rate.

So what persuades Jackendoff of these claims? In both cases, an appeal to `phenomenology' (pp. 15 and 16); indeed, he introduces (5) by saying, "If we consider the phenomenology of attention, ..." (p. 16). And my question is simply this: how does he know that attention is as it seems phenomenologically? Perhaps it merely seems to be that way! (Dennett uses this basically Kantian distinction between even appearances as they appear and appearances as they actually are to great effect in [1991].)

I don't want to belabour these putative problems with Jackendoff's specific claims. If his claims face a problem, so do my objections and putative counter-examples. The problem is this: What could settle the issue between us? Facing Jackendoff's intuitions and intuition-pumps with other intuitions and intuition-pumps may be good fun but we need more than that to get anywhere here.

Jackendoff tends to support his claims by two kinds of appeal. When he is dealing with the relationship of language to thought, he appeals to commonsense observations from everyday life. When he is dealing with consciousness and what does and does not `appear' in consciousness, he appeals to introspection, phenomenology: what we observe or think we observe in ourselves. We need not consider the first kind of appeal here, but we do need to consider the second, how something appears introspectively. It seems to me that many of Jackendoff's claims and intuitions on this issue, and equally my counter-claims and counter-intuitions, face a big problem. Once we are aware of something in some way or other (by inference, by presentation in a linguistic or visual or some other form, by reading about it, by imagining it, whatever) -- once we are aware of something in some way, how do we know whether that thing appears in consciousness or is just known in some other way?

A big problem. Where might we find a solution? Only, I think, in a well-worked out theory of consciousness -- a theory of consciousness, something that can give us a principled means to determine when something is transparent to consciousness, translucent to consciousness, vaguely conscious, the centre of focused consciousness, out of consciousness, in consciousness only via being expressed in a linguistic or visual form, and so on and so forth. Only then will we have more than blank appeals to introspection to guide us.

2. How far does Jackendoff's theory of consciousness take us?

Our next question, then, is this: How far does Jackendoff's theory of consciousness take us? The answer, I think, is: Not very far. Indeed, it is not clear that he is even talking about the same thing on all occasions of use of `consciousness'. Most of the time, he seems to have introspective consciousness and/or consciousness of self in mind (I argue in [1994] that these are different from one another), but sometimes he seems to have consciousness of the world in mind, consciousness of the sort that even quite simple animals have.

That Jackendoff may have the latter in mind, or partly in mind, some of the time is suggested by (5), the claim that we can pay attention only to something of which we are conscious. Taken as a claim about introspective awareness, this claim, as we saw earlier, is not very plausible. As a claim about simple awareness of things around us, however, it becomes more plausible: how could we pay attention to something outside us and not be conscious of it? Attention is just one form of this kind of consciousness. So one wonders if some of the appeal of (5) for Jackendoff might not reside in an implicit conflation of the two kinds of consciousness.

The same notion of consciousness as simple awareness of the world seems to be at work in an example on p. 15, the example of lying on a beach watching people go by. Though there is nothing here that is hard to process, "you're conscious of all of it", as Jackendoff says. All of what? Jackendoff himself tells us. You are not conscious of yourself and your states; you are conscious of the world around you.

This also seems to be the notion of consciousness at work when he considers the old idea that "we are only conscious of things that are hard to process" (p. 14). When we are aware of something that is hard to process, it is that hard-to-process item that we are aware of, not our perception of it or our thinking about it or anything else internal to us. Giving one's full attention, full consciousness, etc., to something is one thing, and being aware of thus being conscious or paying attention is another.

The central notion of Jackendoff's paper, the notion of "appearing in consciousness", is at least ambiguous on the same point. Is it enough for something to appear in consciousness that it is represented in me, or must I also be aware of the representation, be aware that it is represented in me? I think Jackendoff has the latter in mind, but it is hard to be sure.

On the other hand, most of the time Jackendoff clearly has self-consciousness in mind, indeed the form of it found in introspection. He has this sort of consciousness in mind when he says that both sensory information and cognitive states of the central system such as thoughts are unconscious. He does not mean that they provide no consciousness of the world; he means that we are not conscious of them (p. 9). Likewise, this is the sort of consciousness that he has in mind when he says that something is "inaccessible to consciousness" (p. 9). He does not mean that it is not representing something in the world; he means that we cannot introspect it. Likewise, introspectibility has to be the notion of consciousness at work when he says that "thought per se is never conscious" (p. 7).

The distinction between consciousness of the world and introspective consciousness of oneself and one's representations (and other psychological states if any) is very important. As Dretske (1995) has recently argued, and very cogently, only confusion results from running the two together. If the matter is not as clear as it could be in Jackendoff, that may suggest that he does not have much of a theory of consciousness. And I am not sure that he does; in fact, I suspect that I have laid out most of what he offers in the remarks above: the image of consciousness as intermediary, the notion of valuation (which does not play a major role int the paper before us), the intuitive notion of introspection, the claims about the relationship of language to what can appear in (introspective) consciousness, and the observations about the relationship of consciousness and attention. This collection of ideas pretty much seems to exhaust Jackendoff's contribution to our understanding of consciousness.

The crucial notion is the notion of appearing in introspective consciousness. What is it for something to appear in introspective consciousness? Here are some candidate necessary conditions. It is: A matter of being able to report on it. A certain kind of short-term memory. Informational access to the item of a kind that no one else has. (And no one else could have?) Now, how does consciousness relate to these candidate conditions. Are all of them necessary for it to exist? Some of them? Other things, too? And are these or any subgroup of them also sufficient? If not, what more is needed? We need to be able to answer questions like these to resolve what I called the big problem earlier, the lack of any way to determine when something appears introspectively, when it does not, and perhaps even when it does not but appears to. Jackendoff does not offer much to help us.

Against this charge, one might look to his work on attention and focus. Jackendoff certainly makes some interesting claims about these things. Unfortunately, they cannot help us with the problem of delineating the nature and boundaries of consciousness, because in developing his thoughts on them, Jackendoff starts from and takes for granted just the intuitive sense of being in consciousness that is causing the trouble. I cannot find anything further in his analysis, anything that elucidates this intuitive sense or grounds it in something better.

3. What else do we need for a theory of consciousness?

It would be impossible to lay out anything like a complete list of the requisites for a theory of consciousness here -- even if I knew them, which I do not. So I will just sketch a few lines of thought.

Clearly a theory of attention will be part of any completed theory of consciousness, though we can be conscious of things in the world and we can probably also be introspectively conscious of things in ourselves to which we are nevertheless not paying attention. A theory of attention, in turn, will have close links to a theory of focus: of focusing on something, concentrating consciousness and cognitive resources on it, etc. Examples are things like trying as hard as one can to get a clear picture of a problem, of focusing everything on making the right decision, of bringing everything one can think of to bear on a situation. Any complete theory of consciousness must provide an account of activities like these.

Next, a theory of focus will be part of a general theory of mental unity. One species of mental unity that is closely linked to focusing is the unity of consciousness. Indeed, at first blush, focusing on something seems simply to be a matter of the thing becoming a special object of unified consciousness. (It is also to bring unified cognitive resources to bear on it, a point to which we will return.) The question with which Jackendoff is centrally concerned, namely, what appears in consciousness (p. 4), will then become the question of what it is for something to be become part of unified consciousness. So let us look briefly at mental unity.

There are many kinds of mental unity. Indeed, we will shortly distinguish four of them. Since we have been considering focusing on something, let us start with what we might call unity of focus. To see what unity of focus is like, start with the better known unity of consciousness. We might define the latter as follows:

The unity of consciousness (UC) =df. (i) a representing in which (ii) a number of representations and/or objects of representation are combined in such a way that to be aware of any of the represented objects is also to be aware of other represented objects, all as the object of a single representation.

Actually, this is only one form of UC. It is a matter of being aware of a whole group of objects of representation as the object of a single representation. Another consists of being aware of oneself as the common subject of those representations; as Kant (1781: A350) put it, we are aware of the thing that does the representing as the single "common subject" of a multitude of representational states.

Clearly UC in both it's forms is an important part of what we call consciousness. I discuss them further in (1994: Ch. 3). The unity on the volitional side that I call unity of focus may be even more central, however. It consists of the focusing of intentional resources on things, problems, courses of action, etc. We can focus on a number of considerations at the same time and weigh up their implications. We can focus on a number of alternative courses of action at the same time and assess them against one another. We can bring these considerations together to form an intention and choose a course of action. And we can then focus our resources on carrying out that course of action, against obstacles, conflicting desires, and so forth.

Mental unity takes more forms than these two -- or three, if one considers unified consciousness of objects of representations to be a different kind of mental unity from unified consciousness of the representations themselves and of oneself as the common subject of them. In addition to UC and unity of focus, there is what we might call unity of cognition: to form representations of objects, we must bring perceptual, conceptual, memory, linguistic, evaluative, reasoning, affective, volitional, and other capabilities to bear, and in a highly unified way. (We might think of the problem of how we manage to do this as the binding problem writ large.) And at the other end, there is also an important unity of action: the unity achieved by coordinating mouth, limbs, digits, torso, facial expressions, orientation and focus of eyes, and so on and so forth, to produce coordinated actions all of whose parts are in the service of a single common objective.

The unity of action might not be important to a theory of consciousness but all the other kinds of mental unity clearly are. Thus, Jackendoff needs a theory of the unity of consciousness and of focus and of cognition. He needs it because for something to "appear in consciousness" will be for it to become part of such a unified array of representations; this will be the result in turn of a certain application of a unified array of cognitive capabilities. To be able to decide when something is and is not in the array, we must know what it is for something to cross the boundary. To know that, we must know what these kinds of unity are like and what the conditions are for something becoming part of such an array.

To decide whether we are aware of intermediate-level representations (phonological representations in language, 2 1/2D representations in vision) and nothing else, we need a theory of consciousness along these or closely related lines. Similarly for Jackendoff's other claims about consciousness, thought, language, attention, etc.: to assess them with any confidence, we need a theory of consciousness.


Brook, A. 1994. Kant and the Mind. Cambridge University Press.

Dennett, D. 1991. Consciousness Explained. New York: Little, Brown.

Dretske, F. 1995. Naturalizing the Mind. Bradford Books/MIT Press.

Jackendoff, R. 1987. Consciousness and the Computational Mind. Bradford Books/MIT Press.

-----------------. 1996. `How Language Helps Us Think'. This journal.

Kant, I. 1781. Critique of Pure Reason. Trans. Norman Kemp Smith as: Immanuel Kant's Critique of Pure Reason. London: Macmillan, 1963.

Wittgenstein, L. 1953. Philosophical Investigations. Oxford: Basil Blackwell Publishers.