Brie
Gertler
The Narrow Mind
Draft!!
Where does the mind end and the ‘external’ world begin? Some philosophers believe that the mind is contained within the subject’s body, where ‘body’ refers to the brain or nervous system. They maintain that the mind is in this sense narrow. Others, including proponents of ‘embodied cognition’, claim that the mind is wide in that physical features of the world beyond the subject’s body partially constitute her mind. My central goal is to show that the mind—in the sense of ‘mind’ salient for a philosophical conception of the self—is narrower than most philosophers believe. Unsurprisingly, my position is at odds with ‘wide’ conceptions of the mind. But it also conflicts with most ‘narrow’ construals of the mind, for it is even more restrictive than most of these.[1]
My argument will employ what is probably the strongest defense of the wide conception of the mind, that given by Andy Clark and David Chalmers.[2] Clark and Chalmers (‘C&C’ hereafter) describe a relatively mundane situation, one which is undeniably possible. They conclude from this case that items external to the body can contain our mental states; hence, ‘the mind extends into the world’ (12). I think that C&C’s inventive case has the opposite result: it shows that the mind is narrower than we may have believed, rather than wider.
While the debate about the breadth of the mind is usually conducted within a materialist framework, it is worth noting that non-materialists can also use differ as to the extent of the mind. For instance, a dualist who thinks that the mental causally supervenes on the physical could think that the mind is narrow in that its physical supervenience base lies within the body. Alternatively, such a dualist could endorse a ‘wide’ conception, maintaining that the mental’s causal supervenience base lies partly outside the subject’s body. For purposes of this paper, I remain neutral about the debate over materialism. I will argue against the view that a certain kind of physical item partially constitutes the mind, but this argument has no direct implications for the truth of materialism generally. As much as possible, I will avoid such larger ontological issues in this discussion.
1. The Clark-Chalmers argument
C&C’s argument has two main components. The first aims to show that processing systems outside the brain are part of the subject’s mind: for instance, a calculator which a subject keeps clipped to her at all times and consistently uses may be part of her cognitive processing system. Since my chief concern is with particular beliefs and desires, I will focus on the second and more contentious component of C&C’s argument, which aims to show that particular beliefs can lie outside the subject’s brain.
C&C make their case by describing Otto, an Alzheimer’s sufferer. Otto routinely records in his notebook useful information of the sort that most of us would easily commit to memory, and consults the notebook when he needs the information to guide action. For instance, on a trip to the Museum of Modern Art, Otto consults the notebook frequently; his notes remind him that he is going to the museum, that the museum is located on 53rd Street, etc.
Here is my reconstruction of C&C’s argument.
(1) The records in Otto’s notebook play the same causal-explanatory role (vis-à-vis Otto’s behavior, occurrent states, etc.) as standing beliefs within his brain.
(2) Standing beliefs within Otto’s brain are mental states of Otto’s.
(3) Anything which plays the same causal-explanatory role (vis-à-vis behavior, occurrent states, etc.) as a mental state of a subject is itself a mental state of that subject.
So,
(4) The records in Otto’s notebook are mental states of Otto’s. (from (1), (2), and (3))
(5) Anything containing some of a subject’s mental states is part of that subject’s mind.
So,
(6) The notebook is part of Otto’s mind. (from (4) and (5))
And hence,
(7) ‘The mind extends into the world.’ (from (6))
The argument is clearly valid. But the view it seeks to establish is deeply problematic. As I will show, the view posits mental states which are explanatorily superfluous; it entails that the mind may extend beyond notebooks to things that we should be very reluctant to count as part of the mind; and it threatens deep intuitions about privileged access and the distinctness of individual minds. In the face of these costs, the proper move is to regard the argument as a reductio of Premise 2. We should accept that Otto’s notebook is as much a part of his mind as his standing beliefs; but rather than elevating the status of Otto’s notebook, we should discount the philosophical significance of Otto’s standing beliefs.
On my alternative, only occurrent states—occurrent sensations, beliefs, desires, etc.—are unqualifiedly, straightforwardly mental. Non-occurrent states are mental only derivatively, in virtue of their potential to become occurrent or to shape occurrent thought. C&C are well aware of the sort of position I will defend: they mention positions of this sort twice, rejecting them as ‘extreme’. I locate the philosophically interesting boundary which defines the mind in a different place than C&C do, but I agree that the boundary they suggest is a valid, if less philosophically revealing, one. My hunch is that they would view the boundary I suggest as valid but less philosophically interesting than their own. So why not accept both boundaries, and treat this as a merely verbal dispute over the term ‘mental’?
The real dispute here concerns which of these—solely occurrent states, or occurrent states plus non-occurrent states, including external factors—constitute the mental in the philosophically salient sense, that is, in the sense that defines the self. This is clearly how C&C regard the issue. ‘What, finally, of the self? Does the extended mind imply an extended self? It seems so. … What this comes to is that Otto himself is best regarded as an extended system, a coupling of biological organism and external resources.’ (18, emphasis in original) They recognize that the stakes here are very high, for their proposal has ‘obvious consequences for philosophical views of the mind … [and] there will also be effects in the moral and social domains.’ (ibid.) And they expect that their proposal will alter how we ‘see ourselves’. (ibid.)
If my assessment is correct, the external factors which C&C take to partially constitute the mind are not part of the self in the core philosophical sense of ‘self’. I will not specify this core sense precisely, but only characterize it as: that which is crucial in explaining behavior and individuating persons, and which is constituted by states to which the subject enjoys a unique—perhaps, uniquely privileged—access. There may be other important notions of the self, but I doubt that any competing notion is as central to issues about individuating minds. In any case, my claim will be that only occurrent states are, strictly speaking, part of the self in this sense.[3]
All parties agree, then, that the boundary at issue delimits the philosophically salient self. So the dispute over the mental’s true domain is not merely verbal; it carries profound consequences for the wide range of philosophical positions that involve the notion of selfhood. In what follows, I outline three unfavorable consequences of C&C’s proposal, consequences to which my alternative is immune. I then sketch a response to their principal objection to my alternative.
2. First objection to C&C: explanatory
relevance
C&C’s
brand of externalism differs from the more familiar brand of externalism
developed by Putnam (1975) and Burge (1979). C&C claim that the mind includes
external items, like Otto’s notebook, which play an active role in the
subject’s cognitive economy. By contrast, the externalism of Putnam and Burge
concerns external items which appear to be inert, as regards the subject’s
behavior. To use a classic example from Putnam, the water in my
surroundings—or, perhaps, the fact that there is water in my
surroundings—partially constitutes my thought content ‘water’. To mark this
difference, C&C label their brand of externalism ‘active’, and the Putnam/Burge
brand ‘passive’.
These
brands of externalism are mutually independent; one may coherently accept
either while rejecting the other, and one may coherently accept both. But
C&C argue that active externalism avoids one pitfall of passive
externalism, namely, that passive externalism includes, within the mental,
states which appear to be irrelevant to explanations of behavior. By contrast,
C&C claim that their active externalism ensures the explanatory relevance
of mental states.
In counterfactual cases where internal structure is
held constant but these [passive] external features are changed, behavior looks
just the same; so internal structure seems to be doing the crucial work. …
[A]ctive externalism is not threatened by any such problem. The external features
in a coupled system play an ineliminable role—if we retain internal structure
but change the external features, behavior may change completely. The external
features here are just as causally relevant as typical internal features of the
brain. (9)
It is a controversial issue whether the problem of explanatory irrelevance is fatal to the Putnam/Burge version of externalism. But explanatory relevance seems particularly important to C&C’s ‘active’ externalism: after all, Otto’s notebook is claimed to contain his beliefs precisely because the records in the notebooks appear to contribute to his behavior in the same way that his beliefs contribute.
However, it appears that these ostensibly ‘active’ external factors play no more role in explaining behavior than the ‘passive’ features of Putnam/Burge externalism. For the argument which C&C use to show the explanatory irrelevance of ‘passive’ external states provides equally good reason to doubt that the notebook states (or other standing states) are irrelevant to explaining behavior.
Suppose that we want to explain Otto’s walking towards 53rd Street. This movement is partially caused by the fact that his notebook says that the museum is on 53rd Street. But this fact causally influences his movement only because Otto consults the notebook and thereby comes to occurrently believe that the museum is on 53rd Street. If Otto had not formed an occurrent belief about the location of the museum, he would not have walked towards 53rd Street. So Otto’s notebook causally influences his actions only by affecting his occurrent beliefs. Reference to Otto’s notebook is, then, perfectly eliminable in an explanation of his action.[4]
If this explanation of Otto’s behavior is accurate, then C&C’s proposal faces precisely the problem which they describe as afflicting the Putnam/Burge brand of externalism. For occurrent beliefs are presumably internal, and so it is not the case that ‘if we retain internal structure [including all occurrent states] but change the external features, behavior may change completely’. So long as the occurrent states which are causally shaped by external factors are held constant, Otto’s behavior will remain unchanged. Removing Otto’s notebook would causally alter the situation, but so would removing lots of other, unquestionably non-mental factors, such as the sidewalk which Otto uses. (Of course, internalists admit that external features causally influence mental states; the debate over externalism concerns whether external features contribute logically to determining the mental.) So the external factors described by C&C do no essential work in explaining Otto’s behavior.
C&C’s proposal escapes this objection only if non-occurrent states result in behavior directly, without the mediation of occurrent states. And they recognize this requirement. They describe Twin Otto, who is indistinguishable from Otto except that he mistakenly wrote in his notebook that the museum was on 51st Street. They note that Twin Otto’s behavior will physically differ from Otto’s: Twin Otto will walk towards 51st Street. They conclude that ‘the relevant external features … have a direct impact on behavior.’ (14) This follows only if there is no relevant internal difference between Otto and Twin Otto. But it seems clear that there is an internal difference here: when he consults the notebook, Twin Otto comes to occurrently believe that the museum is on 51st Street, while in the same circumstance Otto comes to occurrently believe that it’s on 53rd Street. And these occurrent beliefs influence their subsequent walking behavior. After all, Twin Otto could misread his handwriting and, mistakenly taking the notebook to say that the museum is on 53rd Street, walk in the same direction as Otto. This would result from their internal similarity, for both Otto and Twin Otto would occurrently believe, at the relevant moment, that the museum was on 53rd Street. Holding fixed all internal (and, hence, all occurrent) content, there seems to be no difference in physical behavior between Otto and Twin Otto.
The Twin Otto case does not show that standing states can directly cause behavior, without the intervention of an occurrent state. Of course, there may be independent grounds for thinking that standing states can directly cause behavior. But I think that the case of Otto provides some reason to deny this, for if standing states can directly result in behavior, the following case is possible. Otto programs a robot to carry the notebook’s information, e.g., to accompany him to the museum and, when necessary, to remind him of his destination. The robot thus contains some of Otto’s non-occurrent mental states, according to C&C. Now if non-occurrent states can result in behavior directly, the robot’s movements that are caused by the information Otto has programmed into it, would seem to qualify as Otto’s behavior. To avoid this result, we must require that behavior involve movements of Otto’s body. This requirement seems, at least in spirit, to violate C&C’s principle that ‘when it comes to the mind, there’s nothing sacred about skull and skin’ (14). And it would be difficult to formulate a principled requirement on behavior which excluded the robot’s movement but included movements of prosthetic limbs. (For instance, it seems arbitrary to require that a subject’s nervous system somehow contribute to a movement, in order for it to qualify as her behavior.)
Finally, even if we could somehow require that Otto’s behavior must involve his own body, we would end up with strange results. Suppose that the informational states of the robot, programmed by Otto long ago on the basis of his then-occurrent beliefs and desires, lead the robot to hit Otto’s knee, causing it to shoot out. This paradigm example of non-behavioral movement would then qualify as Otto’s behavior, because it was caused by his beliefs and desires. (This case goes a bit beyond anything that C&C consider, but it seems to me a logical extension of the cases they do consider.)
In any case, the external items which C&C countenance as ‘active’ parts of Otto’s mind seem superfluous in explaining his behavior. These items play important causal roles in determining Otto’s occurrent states, but so do other, unquestionably non-mental items in Otto’s environment. Most importantly, no variation in these so-called ‘active’ factors will change Otto’s behavior, so long as his occurrent states are held fixed. Otto’s notebook affects his behavior, but only through affecting his occurrent states. If this is correct, occurrent states seem to avoid the explanatory superfluity which faces standing states.
C&C maintain that Otto’s notebook is part of his mind because the records in the notebook meet the following four conditions on dispositional belief. (1) The notebook is consistently available; (2) its information is readily accessible; (3) Otto automatically endorses information from the notebook; and (4) the information is present in the notebook because it was consciously endorsed in the past.
My second objection to C&C’s argument stems from the possibility of something which meets these conditions but which we should be reluctant to count as part of the subject’s mind. Consider Elsa, an otherwise normal person whose thoughts are being controlled by something outside of her brain. The controlling entity could be a mad scientist or a computer gone haywire; let’s suppose the controller is a Cartesian evil demon. The evil demon causes Elsa to have particular occurrent beliefs at his whim, beliefs that influence her behavior.
Because the demon’s causal control of Elsa’s states is consistent (alas, demons never sleep), the case meets condition (1). It meets condition (2) by virtue of the fact that Elsa never needs to decode the demon’s messages—he directly causes her to instantiate occurrent beliefs as he pleases. And since Elsa is powerless to avoid the beliefs which the demon causes her to instantiate, the case meets condition (3).
Strikingly, the most difficult condition for the demon to meet is also the condition about which C&C are most hesitant. They allow that condition (4) may not be necessary for belief. ‘[P]erhaps one can acquire beliefs through subliminal perception, or through memory tampering’. (17) In any event, an added twist allows the demon case to meet condition (4). Suppose that at an earlier time, Elsa had introspected some beliefs that she considered unpleasant. Wishing to be rid of the beliefs, and mistakenly taking the demon to be a virtuous creature, she willingly relinquished control over her thoughts to him. Now she may never have endorsed the particular beliefs that the demon causes her to instantiate. But she might have granted the demon carte blanche to make decisions about the beliefs she should instantiate. If so, this case satisfies condition (4). (It is similar to a case C&C seem willing to allow, that for one who is ‘unusually computer-reliant, facile with the technology, and trusting’, the mind may extend as far as the Internet.(17)) The trustworthiness of the demon is irrelevant; C&C never claim that the mind extends only to properly functioning or reliable systems, nor is such a claim defensible.
So Elsa’s relationship to the demon meets C&C’s four conditions, and thus mirrors Otto’s relationship to his notebook. Does this show that C&C’s argument implies that evil demons might be part of our minds? And would this be grounds for rejecting the argument? It might appear that the case of Elsa poses no problem for C&C’s argument. According to one response, there are no demons in this world, and any world in which there are demons is a world where the nature of minds is very different from their nature here. I think this response is inadequate, for we’re after a principled, philosophically significant boundary of the mind. That is, Premise 3 of the original argument—that anything which shares a causal-explanatory role with a mental state of a subject is itself a mental state of that subject—is presumably a conceptual claim about what qualifies as mental. It is therefore not limited to our world or to worlds sufficiently similar to ours.
An alternative response is to allow that the demon simply is part of unfortunate Elsa’s mind. This is the answer I’d expect C&C to give. But it does threaten a deep intuition. For suppose that Elsa reads Descartes’ Meditations, and muses on whether a demon controls her thoughts. Surely she is musing on whether she is under the control of something outside of herself, an alien force, a foreign influence. The states that the demon causes her to instantiate, like the states Otto’s notebook causes him to instantiate, are occurrent states. These unquestionably belong to her (and C&C do not claim that these extend beyond her brain). Her musing, then, takes the form of wondering whether there is something outside of her which is causing her to instantiate these occurrent states. The obvious intuition is that the sense of ‘outside’ at work here is relative to Elsa’s mind. And we can accommodate this intuition only by denying that the demon is part of Elsa’s mind.[5]
My second objection to C&C’s view is, then, that it entails that a Cartesian evil demon—which appears to be an outside, foreign, corrupting force—could be part of the subject’s own mind. My proposal sidesteps this consequence. On my proposal, Elsa’s mind is constituted by her occurrent states. The causal sources of those states—demons, mad scientists, notebooks, the Internet—are, strictly speaking, not part of her mind. This makes sense of Elsa’s worry that something outside her own mind is controlling her (occurrent) mental states. It also fits nicely with Descartes’ view that fully grasping one’s own (mental) nature does not require knowing the causal sources of one’s occurrent states. (Some may not regard this latter result as advantageous, of course.)
4. Third objection to C&C: privileged
access
Suppose that Otto has an identical twin, Ingo, who also suffers from Alzheimer’s disease. Otto and Ingo live together in close quarters, and for efficiency they share a single notebook. Each records information in the notebook, and each trusts the other’s inscriptions as much as he trusts his own. We can easily imagine that they are unable to determine which of them has inscribed a given piece of information; what matters is that each is disposed to endorse the information in the notebook, believing that it came from either himself or his twin. Now if the notebook is part of Otto’s mind, and contains his beliefs, then it is also part of Ingo’s mind, and contains his beliefs. Otto and Ingo thus have overlapping minds. That is, they share beliefs—not just in the ordinary sense of sharing belief types, but in the extraordinary sense of sharing belief tokens.
C&C recognize this sort of possibility. ‘In an unusually interdependent couple, one partner’s beliefs will play the same sort of role for the other as the notebook does for Otto.’ (17) But the consequence that two minds can overlap is deeply troubling. Aside from obvious questions it poses about personal identity, it seems to conflict with the widely-held intuition that each of us enjoys privileged access to our own mental states, that is, access that is different, in principle, from others’ access to those states. After all, Otto and Ingo have equal access to the states they share.
Might privileged access be accommodated nonetheless? Acknowledging that Otto and Ingo have equal access to particular beliefs, one could maintain that Otto has privileged access to the fact that he believes that p. For Otto can determine that he believes that p simply by consulting the notebook. By contrast, to determine that Otto believes that p, Ingo must both consult the notebook and determine that the notebook plays the appropriate role in Otto’s cognitive economy.
But this strategy does not work. Imagine that, unbeknownst to Otto, someone replaces his notebook with another, indistinguishable from the original except for one particular. Where Otto’s notebook says that the museum is located on 53rd Street, the substitute says that it’s on 51st Street. After a short while, perhaps an hour or so, Otto’s original notebook is returned to its place. Otto remains unaware of either switch. Because the substitute is in place for too brief a period for its contents to qualify as Otto’s beliefs, it fails to meet C&C’s conditions (1) and (4) for qualifying as Otto’s standing states. Now suppose that, while the substitute was in place, Otto consulted it to determine what he believed about the museum’s location. Insofar as he took the notebook to contain his standing beliefs, he would be mistaken. For he would have mistakenly assumed that this notebook plays the appropriate role in his cognitive economy. While this assumption is implicit, it is precisely the step that Ingo must take, in determining Otto’s beliefs. So Otto’s access to his standing beliefs, through his notebook, hinges on the same assumption as that which constrains Ingo’s access to Otto’s standing beliefs. This is, then, the third objection to C&C’s proposal: it cannot accommodate privileged access, for Otto’s access to his standing beliefs is, in principle, the same as others’ access to them.[6]
Even if they
have equal access to standing beliefs, Otto’s access to his own occurrent
states may yet differ, in principle, from Ingo’s access to them. By consulting
the notebook to determine his standing belief about where the museum is
located, Otto thereby comes to occurrently believe that the museum is on 51st Street. This may
form the basis for an accurate introspective self-ascription, namely, ‘I
occurrently believe that the museum is on 51st Street’. The
intuition that we have privileged access to our own mental states applies most
clearly and directly to occurrent states. As I see it, the Otto case has a
hidden benefit: it both explains and justifies this feature of the intuition
about privileged access. For the records in Otto’s notebook are on a par with
(or just are) his standing beliefs, and others can, in principle, have
equal access to them. If privileged access is salient to personhood, then
C&C’s Otto case provides another reason to take the philosophically salient
boundary of the self as that which divides occurrent beliefs, desires, etc.,
from other states, including standing states.
C&C explain why they reject a proposal like mine. ‘To consistently resist this conclusion [that Otto himself is an extended system], we would have to shrink the self into a mere bundle of occurrent states, severely threatening its deep psychological continuity.’ (18) This is the most difficult challenge for my alternative view to meet.
I can only briefly sketch how we might secure Otto’s psychological continuity, while identifying him with his occurrent states. The approach uses states which are not part of Otto’s mind as the causal ground for his psychological continuity. Imagine an automobile factory that produces several different models of cars. The cars may bear the factory’s unique mark and, thus, their internal features may express their shared origin; or they may be consecutively numbered. But even if they bear no common or related internal marks, they still form a unified, causally salient class, viz., the class of products of that factory. The factory’s causal continuity grounds the causal salience of that class, as products of the same cause (or, perhaps, as products of factory stages which are so related as to qualify as causally continuous). This is so despite the fact that the factory itself does not figure in the class.
A similar picture applies to occurrent states. Suppose that a group of occurrent states have a shared origin. They spring from states, including standing beliefs and non-mental states, of a persisting physical organism (or causally continuous stages thereof). This shared origin may produce occurrent states which share internal features: e.g., one’s occurrent states may share a quality of hopefulness because they spring from physically-based dispositions to be optimistic. And they may have complex rational interrelations, derived from appropriately structured features of their shared origin. But as in the factory case, it is the causal continuity of the shared origin itself which renders the states causally unified. Any shared internal features or coherent rational structure is unnecessary—for instance, it is not present in Otto’s presumably jumbled occurrent states. And the shared origin needn’t be included in the class of occurrent states in order to ground its causal salience, any more than the factory needs to be included in the set of cars to ground its causal salience.
This is just a sketch of a view, with many details to be worked out. I use it only to suggest that ‘psychological continuity’ may be in place even if the self is “a mere bundle of occurrent states”.
Of course, C&C could deny that the boundary of in-principle privileged access is the mark of the mental, and could hold that Elsa is simply mistaken in taking the (evil) source of her occurrent beliefs to lie outside of herself. But the resulting picture of the self is starkly at odds with the notion of the self relevant to classic philosophical problems, the notion which is their explicit target. And invoking explanatory concerns will hurt rather than help their case.
These problems with the conclusion that the mind extends into the world are, I think, serious enough to warrant treating C&C’s argument as a reductio. The guilty premise here is Premise 2, the claim that standing beliefs within Otto’s brain are mental states of Otto’s, in the sense of ‘mental’ that defines the philosophically salient self.
Given their intimate and
significant causal relationships with occurrent mental states, it is not
surprising that things outside the brain, such as Otto’s notebook, may also be
philosophically interesting. But the boundary of the mind salient for
philosophical problems involving the self counts standing states as outside the
mind, even if they are inside the brain. This is a surprising conclusion but,
on balance, it is less costly to intuitions and hence less ultimately
surprising than the claim that items such as notebooks can contain our mental
states.
[1] The exception is Strawson (1997) “The Self”. Journal of Consciousness Studies 4: 405-28
[2] Clark, A., and D. Chalmers. 1998. The extended mind. Analysis 58: 7-19.
[3] It will be noticed that my characterization doesn’t include any moral notions. Perhaps being worthy of moral consideration is central to a philosophical notion of the self. It seems to me that my proposal can accommodate at least some theories about what being worthy of moral consideration consists in, but I will not address these here.
[4] C&C complain that it would be ‘pointlessly complex’ to explain Otto’s action by invoking his internal standing belief that the notebook says where the museum is located, in addition to the external fact that the notebook says it’s on 53rd Street. This complaint targets the view that external states are explanatorily irrelevant but internal states that are otherwise on a par with those external states, viz. standing states, are explanatorily relevant. It does not affect my proposal, which limits explanatory relevance to occurrent states.
[5] This sense of being acted upon by an outside force remains regardless of whether we accept condition (4). For that condition ensures only that the subject acquiesced in being controlled by the notebook or the demon at some previous time. And as C&C rightly point out, historical causes seem insufficiently active to constitute one’s mind at present. In note 5, they suggest that the conditions might be modified to avoid purely historical causes, in part by removing condition (4). But far from ensuring the subject’s activity, eliminating condition (4) would allow the demon case to meet these conditions while Elsa was even more passive, relative to the demon, than she would be if she willingly subjected herself to his control.
[6] I use ‘others’ plural, though I’ve shown only that Ingo’s access is the same as Otto’s. But the Otto and Ingo case could be developed so that any number of persons had equal access to the notebook. (The possibility of a large number of persons sharing standing belief tokens is clear from C&C’s Internet example.)