Externalism and Causality: Simulation and the Prospects for a Reconciliation   

Dōna Warren

 



Abstract

Externalism in the philosophy of mind has been invoked by some philosophers to argue that content-bearing mental states can’t serve as the explananda in genuinely causal explanations of behavior.  In this paper, I demonstrate that such arguments presuppose that psychological explanations are theory-based and that if this theoretical conception of psychological explanation is replaced by the simulation model we remove the source of the apparent tension between externalism and causality and are in a position understand how appeal to content-bearing mental states may be causally explanatory.

Externalism and Causality: Simulation and the Prospects for a Reconciliation*

 

1. Two Guiding Assumptions of Everyday Psychology

 

                Everyday psychology, otherwise and sometimes pejoratively known as ‘folk psychology,’ is comprised of our practices of attributing mental states to each other and of predicting and explaining each other’s behaviors on the basis of the mental states ascribed.  It is guided by two fundamental assumptions.  First, we suppose that these everyday psychological predictions and explanations are causal in character; we assume that the mental states cited in a correct prediction or explanation of someone’s behavior actually caused the behavior predicted or explained.  Second, we assume that at least some of the mental states cited in such explanations are ‘about something,’ or are otherwise representational, a presupposition which inclines us to interpret ascriptions of such mental states as expressing a relation between the subject and some idea, proposition, or sentence which serves as the content of that state.     

                Quite a bit needs filling-in, of course, but the overall structure of everyday psychology is so familiar and ingrained that it seems counterintuitive to suppose that anything could be fundamentally wrong with the explanatory framework.  Recently, however, this picture has been challenged by philosophers who argue that the nature of content-bearing mental states is incompatible with their serving as the explananda in genuinely causal explanations or predictions.  In other words, these philosophers argue that the two basic assumptions which we bring to everyday psychology - the assumption that everyday psychological predictions and explanations are causal and the assumption that such predictions and explanations appeal to the subject’s content-bearing mental states - are ultimately inconsistent.   In the second section of this paper, I will critically examine the argument for this inconsistency and show that it depends upon the  background assumption that psychological explanations involve appeal to law-like generalizations.  In the third section, I will argue that the apparent tension between content and causality may be resolved by abandoning the generalization model of psychological explanation in favor of a model based on imaginative identification.

 

2. Externalism and Causality: the Argument for the Inconsistency

 

2.1 Externalism

 

                The argument for the causal inefficacy of content-bearing mental states rests upon the thought experiments of Putnam and Burge, which give us reason to believe that representational mental states are sensitive to external factors in a way which allows them to vary across physically identical individuals (Putnam, 1975; Burge, 1986).[1]  Suppose, for example, that there is another world called ‘Twin Earth’ which is identical to Earth except for the absence of water, H2O, and the presence of a substance, XYZ, that is phenomenologically indistinguishable from water under normal circumstances.  When a linguistically proficient inhabitant of Earth, Oscar 1, speaks sincerely about the substance that fills the lakes and rivers around him, saying ‘Water is good at quenching thirst,’ we may correctly maintain that Oscar 1 is saying that water is good at quenching thirst, and since he’s sincere when he says this, believing what he says, we may conclude that Oscar 1 believes that water is good at quenching thirst.  Equivalently, we may conclude that Oscar 1 has a belief with the content ‘water is good at quenching thirst.’ 

                It’s otherwise when Oscar 2, a linguistically proficient inhabitant of Twin Earth, speaks sincerely about the substance that fills the lakes and rivers around him, saying ‘Water is good at quenching thirst.’  Here, the process of indirect quotation by homophonic translation breaks down.  We can’t maintain that Oscar 2 is saying that water is good at quenching thirst, because there is no water, H2O, on Twin Earth and he hasn’t acquired the concept of water.  We must, instead, coin a term, say ‘twater,’ to refer to the Twin Earthian substance XYZ and to translate the Twin Earthian term ‘water.’  When Oscar 2 says ‘Water is good at quenching thirst,’ he’s saying that twater is good at quenching thirst, and since he’s sincere when he says this, believing what he says, we may conclude that Oscar 2 believes that twater is good at quenching thirst.  Equivalently, we may conclude that Oscar 2 has a belief with the content ‘twater is good at quenching thirst.’

                On the reasonable supposition that content is essential to belief, the fact that Oscar 1 believes that water is good at quenching thirst but does not believe that twater is good at quenching thirst, and the fact that Oscar 2 believes that twater is good at quenching thirst but does not believe that water is good at quenching thirst, lead us to conclude that Oscar 1 and Oscar 2 have different beliefs.  None of this is impugned by the additional assumption that Oscar 1 and Oscar 2 are molecular duplicates of each other, and so it seems that physiologically identical individuals may have different content-bearing mental states.[2]  This is psychological externalism.

                Reactions to externalism have been mixed, and include attempts to articulate a notion of content which is immune to remote environmental influences and which will consequently be shared by physiological duplicates (Fodor, 1982; Loar, 1991).  At the very least, the efforts to develop a theory of so-called ‘narrow content’ have been disappointing and Owens has convincingly argued that despite their initial promise, all attempts to develop a notion of narrow content based on the conceptual role of mental states are doomed to failure; when conceptual role is spelled out in a way which guarantees that physical duplicates will share states individuated according to conceptual role, the states loose any claim to be content-individuated (Owens, 1987).  For the purpose of this paper, we will assume that externalism is correct and that the only viable notion of content is wide content of the sort which may differ across physical duplicates.  Such an assumption will serve two purposes.  First, it will allow us to avoid embroiling ourselves in a discussion of narrow content which would be beyond the scope of this paper.  Second, and more importantly, this is the position which has cast a shadow over the putative causality of psychological explanation, and if we can argue that psychological explanation may be causal even in the face of it, any weaker position with respect to externalism would only stand causality on even surer ground.

 

2.2 The Argument

 

                Fodor has argued that externalism is incompatible with the causality of psychological explanation by taking it for granted, at least provisionally, that psychological explanation is causal and by concluding that this assumption entails that the psychological states to which such explanations appeal cannot differ across physical duplicates. 

                If psychological explanation is causal, Fodor maintains, then one of the following arguments must get the matter right: either (1) Fodor and his hypothetical molecular duplicate engage in actual and counterfactual behaviors which are the same in all relevant respects, leading us to infer that Fodor and his molecular duplicate share mental states which have the same causal powers and so leading us to conclude that the mental states of Fodor and his twin should be typed the same for psychological explanatory purposes, or (2) Fodor and his molecular duplicate engage in actual and counterfactual behaviors which differ in relevant respects, leading us to infer that Fodor and his molecular duplicate enjoy mental states which have different causal powers and so leading us to conclude that the mental states of Fodor and his twin should be should be typed differently for psychological explanatory purposes (Fodor, 1991, p. 5). 

                It’s true, Fodor grants, that molecular duplicates can engage in behaviors which differ in their intentional properties (Oscar 1 can drill for water, for instance, whereas Oscar 2 cannot), but mental states don’t differ in their causal powers in virtue of their being responsible for purely intentional behavioral difference.  Thus, depending upon how you look at it, either the first premise in the second argument is wrong because Fodor and his molecular duplicate don’t engage in behaviors which differ in any relevant respect (intentional differences just aren’t relevant) or, granting the relevance of intentional behavior differences, such differences don’t support the inference to the proposition that Fodor and his molecular duplicate have mental states with different causal powers.[3]  The second argument is therefore either unsound or invalid, and so the first argument must be right: the behaviors of Fodor and his twin are identical in all relevant respects, from which we may infer that Fodor’s mental states and those of his twin have the same causal powers, and from this we may conclude that the mental states of Fodor and his twin should be typed the same for the purposes of psychological explanation.  Therefore, on the assumption that psychological explanation is causal, Fodor purports to have proven that the psychological states to which such explanations appeal cannot differ across physical duplicates, or equivalently, if externalism is true, then psychological explanation cannot be causal.

                Of course, not everyone agrees.  Burge, for one, believes that externalism is compatible with causal explanation in psychology, and argues that the mental states of molecular duplicates can have relevantly different causal powers, thereby justifying their externalistic individuation even on the assumption that psychological explanation is causal (Burge, 1989).  On the face of it, however, Fodor’s position certainly enjoys more plausibility in this respect, because it’s hard to imagine how two physically identical individuals could differ in their causal powers. Consider two identical metal spheres, for instance, or two identical blocks of wood.  How could such things differ in what they are able to do?  How could such things differ in their causal powers?  Molecular duplicates such as Oscar 1 and Oscar 2 are just as alike as two identical metal spheres, and so it certainly seems as though one would be able to do exactly what the other can do, and no more (this point is also made by Owens, 1993, p. 253).  

                Regardless of how this contest turns how, however, it’s interesting to note that the debate regarding the compatibility of externalism and causality has focused around arguments for and against the ability of molecular duplicates to engage in relevantly different behaviors, or, equivalently here, for and against the mental states of molecular duplicates differing in their causal powers.  The presupposition that causal explanation in psychology would require mental states individuated according to causal powers has gone unquestioned, and we might wonder why this assumption has proven so uniformly seductive.[4]

 

2.3 The Theory Theory

 

                I maintain that the allure of individuation according to causal powers stems from the assumption that mentalistic explanation conforms to the deductive-nomological model that governs the physical sciences, according to which specific causal explanations are generated and justified by appealing to general backing laws embedded in a larger theory of the science.  We might, for instance, ask a chemist what caused formula F  to explode.  In answer, we might be informed that formula F was composed of chemical C, that chemical C causes an explosion when mixed with chemical D, and that a careless lab assistant had added chemical D to formula F.  Such an explanation would invoke a chemical law of the form ‘If chemical C is mixed with chemical D, then an explosion will result,’ together with the additional assumptions that formula F was composed of chemical C and that chemical D was added to formula F.

                The position that psychological explanations rest upon some such system of law-like generalizations is called the ‘theoretical assumption,’ or the ‘theory theory.’  The theory theory has the benefit of unifying our account of everyday psychology with our accounts of other everyday abilities.  It’s reasonable to suppose, for instance, that when a speaker judges the grammaticality of a sentence in her native language, she deploys, perhaps rapidly and unconsciously, a grammatical theory of the language (Davies and Stone, 1995).  Similarly, advocates of the theory theory maintain that we ascribe mental states and predict and explain behavior by deploying some internalized, and probably tacit, psychological theory.[5]  Somehow or other, normal adults have acquired a theory about the mind which relates stimulus inputs to mental states, mental states to each other, and mental states to behavioral outputs. Such a theory would incorporate law-like generalizations of the forms ‘If a person, P, is subject to environmental stimuli S1, S2, S3 and so on, then P will enter into the mental states M1, M2, M3, and so on,’ ‘If a person, P, is in mental states M1, M2, M3, and so on, then P will enter into mental states M4, M5, M6 and so on,’ and ‘If a person P is in mental states M4, M5, M6 and so on, then P will engage in the behaviors B1, B2, B3, and so on.’ 

                According to this model of everyday psychology, when ascribing a mental state to an individual we might note the stimuli to which that individual is subject and invoke the generalizations which link such stimuli to mental states.  When predicting an individual’s behavior, we might apply the generalizations which link the mental states previously ascribed to action, and when explaining an individual’s behavior, we might hypothesize a set of beliefs which, if we ascribed them to the subject, would have us allowed us to predict the behavior we wish to explain.  If, for example, we want to ascribe a mental state to Professor Jones, we might note that Jones has just read a nasty response to his paper, (subconsciously) invoke a generalization of the form ‘When individuals receive harsh criticism, they believe that they have been attacked,’ and so maintain that Professor Jones believes that he has been attacked.  If we want to predict what Professor Jones will do after he reads the scathing response to his paper, we might invoke a generalization of the form ‘When individuals believe that they have been attacked, they engage in defensive behavior,’ and predict that Jones will submit a rejoinder to the journal.  If we want to explain an act that Jones has already done, say removing that issue of the journal from the departmental library, we might maintain that he did so because he wished to preserve his reputation among his colleagues, on the grounds that the ascription of that mental state would have allowed us to predict Jones’ behavior, supplemented with the generalization ‘When an individual wishes to preserve his reputation among his colleagues, he may abscond with materials in which criticisms of his work are set forth.’

                In such a nomological picture, our understanding and assessment of causal claims are inextricably intertwined with our understanding and assessment of the theory that stands behind these claims.  This pushes us toward the individuation of explanatory states according to causal powers because adequate explanatory theories require the exclusion of false generalizations and the inclusion of true ones, whereas individuating in a way that types causally different things the same, or types causally identical things differently, would either yield false generalizations or miss true generalizations, respectively. 

                To see why this is so, let’s return to a consideration of metal spheres and try to construct theory, or a set of law-like generalizations, to explain the connection between the placement of these spheres on the platform of a scale and the movement of the indicator dial.  When sphere A is placed on the platform, for instance, the indicator points to ‘1.5,’ but when sphere B is placed on the platform, the indicator points to ‘2.0.’  Presumably, weight is the only property of the spheres that is causally relevant to explaining the behavior of the indicator, and so we may say that spheres are causally identical for these purposes if and only if they weigh the same amount.  If we type the spheres according to their weight, the relevant explanatory generalizations are rapidly forthcoming: ‘If a sphere weighing 1.5 lbs. is placed on the scale, then the indicator will point to ‘1.5,’’ for example.  Suppose, however, that we’re stricken with the constant quirk of typing according to color, and suppose that sphere A is red, leading us to opt for the tentative generalization ‘If a red sphere is placed on the scale, then the indicator will point to ‘1.5.’’  But of course this is wrong.  A red sphere weighing 3 lbs. would not cause the indicator to point to ‘1.5.’  Our color-typing would lead us to class causally different things the same, thereby yielding a false generalization.  Moreover, our theory would be unable to account for the fact sphere C, a blue sphere weighing 1.5 lbs., behaves in all relevant respects identically to red sphere A: they both make the indicator point to ‘1.5.’  Thus, our color-typing suffers from the additional vice of classifying causally identical things differently, thereby rendering us incapable of formulating true generalizations, in this case the generalization which subsumes both sphere A and sphere C.   

                To review, explanation by appealing to theory requires the individuation of explanatory states according to causal powers, because only then will the theory invoked be uncorrupted by the inclusion of false generalizations or vitiated by the exclusion of true ones.  The implications of this for the theory theory of psychological explanation are immediate.  Just as we’ve observed that we mustn’t type spheres A and C differently, on pain of overlooking the generalizations which subsume them both, Fodor observes that a psychologist must not allow water thoughts and twater thoughts to be different states with the same causal powers, on pain of missing the generalizations which subsume both Oscars 1 and 2. ‘Good taxonomy,’ Fodor reminds us, ‘is about not missing generalizations,’ (Fodor, 1991, p. 25.)  Thus, to the extent that we think of psychological explanations as akin to the theoretical explanations offered in nonmentalistic sciences, individuation of mental states according to causal powers will appear virtually irresistible.

                We’ve observed that Fodor’s argument for the incompatibly of externalism and causality depends upon the assumption that causal explanation in psychology requires mental states individuated according to causal powers, and we’ve noted that that this assumption gains its plausibility from the presupposition that psychology conforms to the theory theory.  This puts us in the position to see an interesting possibility: perhaps, by abandoning the theory theory, we can skirt commitment to individuation according to causal powers, and without this assumption, Fodor’s argument for the incompatibility of externalism and causality collapses.  Perhaps, in other words, the perceived tension between externalism and causality derives more from our intuitions about theory construction than from the demands of causation per se, and perhaps some other, nontheoretical model of psychological explanation can reconcile causality with externalism.

 

3. Simulation and the Prospects for a Reconciliation

 

3.1 Simulation

 

                The simulation alternative provides us with an account of mental state ascription and the prediction and explanation of behavior that isn’t dependent upon the assumption of a tacit mentalistic theory.  According to the simulation alternative, we ascribe mental states to an individual, and predict and explain that person’s behavior, by imagining ourselves in her position and determining what mental states we would have or what actions we would perform if we were her.  More specifically, when ascribing mental states, we imaginatively take on the individual’s perceptual and known epistemic circumstances, feed those imaginary inputs into our own cognitive systems, and determine what thoughts and emotions would be generated as a result.  When predicting an individual’s behavior, we imagine ourselves in that person’s situation with that person’s mental states and we determine what actions we would perform. When explaining an individual’s behavior, we attempt to find a plausible set of mental states such that given those hypothetical mental states as input into our own cognitive systems, those systems generate an off-line decision to engage in that behavior (this is roughly the exposition of simulation advanced by Goldman, 1989). 

                For example, again, if we want to ascribe a mental state to Professor Jones, we might note that Jones has just read a nasty response to his paper, imagine what thoughts we would have if we had just read a nasty response to our paper, note that we would believe that we had been attacked, and so maintain that Professor Jones believes that he has been attacked.  If we want to predict what Jones will do after he reads the scathing response to his paper, we would imagine ourselves in Professor Jones’ position, imaginatively taking on his beliefs, desires and character traits, and note that, given those adjustments, we would submit a rejoinder to the journal.  That, we predict, is what Jones will do.  Similarly, if we want to explain why Jones removed from the departmental library the offending issue of the journal, we would put ourselves in Jones’ place and ask what mental states would have motivated us to behave in that way.  Noting that given the adjustments to Jones’ position, we would have removed the journal in order to preserve our reputation among our colleagues, we explain Jones’ behavior by attributing that desire to him.  Of course, we may not be introspectively aware that this is what we’re doing when we ascribe mental states or predict and explain behavior because the entire process may occur automatically and on the sub-personal level.  The important point is that, introspectively accessible or not, the simulation does not depend upon the conscious or unconscious deployment of a psychological theory.[6]  

                The difference between theory theory and the simulation alternative has been illustrated by an analogy with two ways in which one might explain or predict the behavior of an airplane (Gordon, 1992). On the one hand, one could explain or predict a plane’s behavior by deploying some aerodynamic theory, appealing to the relevant generalizations, plugging in the appropriate values, doing the necessarily calculations, and so forth.  This would be analogous to the way in which theory theory would have us explain or predict someone’s behavior; instead of aerodynamic theory, we would  deploy our internalized psychological theory; instead of calculations, we’d run some sort of derivations, and all of this might well go on below the level of conscious awareness.  On the other hand, we might explain or predict a plane’s behavior by constructing a model of the plane, placing the plane in an appropriate environment (maybe a wind tunnel), manipulating the model to conform as completely as possible to the real plane’s situation, and seeing how the model behaves. This would be analogous to the way in which the simulation alternative would have us explain or predict someone’s behavior; instead of a model plane, we’d use the resources of our own mind to model someone else’s mind; instead of manipulating the model to conform to the conditions of the real plane, we’d feed hypothetical mental states into our own cognitive mechanisms, and once again, all of this might occur without our direct knowledge.

                It would be beyond the scope of this paper to argue that the simulation alternative is the correct account of psychological explanation; the debate between advocates of simulation and defenders of the theory theory is ongoing.  Our purpose here is only to demonstrate that the source of the apparent tension between externalism and causality lies in the widely taken-for-granted theoretical model of psychological explanation and to show that this tension may be resolved by adopting the simulation alternative, an account of psychological explanation which is simultaneously compatible with externalism and causality.  We must now proceed to the latter task.

 

3.2 Simulation and Externalism

 

                We will begin by examining the ability of simulation to accommodate the externalistic arguments; we will examine, in other words, whether or not someone employing simulation can ascribe to Oscar 1 the belief that water is good at quenching thirst while ascribing to Oscar 2 the belief that twater is good at quenching thirst. 

                At first glance, externalism would seem to pose no particular problem for the simulation alternative.  After all, externalism is concerned with the influence of context on mental state ascription, and since the process of simulating an individual involves imagining ourselves in that person’s environment, the notion of context plays a crucial role in the very idea of simulation.  Of course the context makes a difference.  Alvin Goldman seems to share this assurance that simulation can allow the external world to affect mental state attributions, writing

 

 Although my discussion centers on the psychological dimensions involved in content attributions, it by no means precludes an important role for the external world, especially causal relations with the external world, in the choice of semantic assignments to thoughts.  In deciding what is the referent of an imputed thought, in particular, it seems clear that the interpreter takes into account the thought’s causal history. Similarly, it is plausible to suppose that imputation of other semantic dimensions of thought involves mind-world connections.  These are plausibly part of the conceptual background with which the interpreter operates.

(Goldman, 1989, p. 180; Goldman’s emphasis)

 

                According to this complacently compatiblist approach, imagining ourselves in the subject’s context presumably includes taking the subject’s environment into account, with all of the externalistic consequences which that entails.  If we’re simulating someone on Earth, for example, we would imagine ourselves looking at water, hearing about water, believing that water is hydrating and so forth; such inputs would result in the generation of off-line beliefs about water, such as the belief that water is good at quenching thirst.  If we’re simulating someone on Twin Earth, on the other hand, we would imagine ourselves looking at twater, hearing about twater, believing that twater is hydrating and so forth; such inputs would result in the generation of off-line beliefs about twater, such as the belief that twater is good at quenching thirst.  The simulation procedure, in other words, involves providing our off-line system with pretend inputs which are already externalistically determined (e.g. ‘believing that twater is hydrating’), and given inputs with broad content, the mental states generated are bound to be broad as well.[7]

                On second look, however, this approach faces a possible problem.  How, specifically, do we take the external world into account in the course of our simulation and consequent belief ascription?  How do we ensure that broad mental states serve as input to the simulation, and that broad mental states are generated as output?  We can easily imagine an objector worrying that so far simulation has been described as an account of the procedure governing the prediction and explanation of behavior and the ascription of mental states.  In order to ascribe a mental state, however, it seems that we must possess the concept of that mental state.  Can simulation enable us to master mentalistic concepts in addition to enabling us to wield them in ascription?  And if so, will the mentalistic concepts mastered be compatible with externalism?  If we can only master and deploy externalistic mental concepts by invoking strands of a theory about the mind, then simulation alone, unsullied by ad hoc appeals to the theory theory, will be unable to take the external world properly into account.

                Contrary to the general direction of our objector’s worries, I believe that the question of mentalistic concept acquisition is a red-herring in this context.  Although it is undoubtedly true that simulation is compatible with externalism if it is sufficient to provide us with mastery of full-blown and environmentally-sensitive mentalistic concepts, simulation may be compatible with externalism even if it is unable to provide with complete command of mentalistic notions. This is because the present focus of concern shines specifically on the externalistic element of the concept of belief and not on the concept of belief as a whole.  The question at hand is ‘Can someone employing simulation account for the environmental sensitivity of belief ascriptions without employing strands of a theory about the mind?,’ not ‘Can someone employing simulation master the concept of belief through simulation alone?’  In order to see that the first question need not wait upon an answer to the second, assume for the moment (what I’m not yet prepared to actually grant) that simulation is insufficient to provide us with a robust concept of belief; assume, for instance, that a full understanding of the representational nature of belief cannot be obtained through a process of simulation.  Such insufficiency would threaten the ability of simulation to accommodate externalism only if the distinctively externalistic elements of belief could not be provided by simulation, or (equivalently) only if simulation could not properly and independently account for the dependence belief ascriptions on environmental factors.  Thus, to defend the ability of simulation to accommodate externalism, we need only argue that simulation is sufficient to capture the specifically externalistic influences on belief ascription, regardless of how the full concept of belief is acquired.  I propose to demonstrate that simulation can in fact capture these externalistic influences by showing that environmentally-sensitive belief ascriptions can be generated by someone deploying simulation even without any appeal on her part to a well-developed concept of belief.  This being the case, simulation is sufficient to capture the dependence of belief upon externalistic influences, regardless of how the complete concept is eventually mastered.

                As Gordon observes in ‘Simulation without Introspection or Inference from Me to You,’ most accounts of simulation have assumed that it must involve the generation of off-line mental states, the ascription of such mental states to ourselves, presumably through the deployment of some special introspective faculty, and finally the ascription of such mental states to the subject, presumably through an analogical inference to the effect that the subject’s mental states are similar to what ours would be (Gordon, 1995, p. 53).  In the case of Professor Jones, for instance, I would imagine that I, Dōna Warren, am an arrogant and irascible male physics professor at an Ivy League university who has just read a scathing response to my paper on radioactive decay.  I would note, via introspection, that under those conditions I would believe that I had been unjustly attacked by inferior scholars and I would infer, via an analogical inference, that Professor Jones actually is as I would be under those counterfactual circumstances: specifically, that he believes he has been unfairly attacked by inferior scholars.

                Such an introspective / analogical model of simulation poses some obvious problems.  First, it’s unclear what theory of personal identity would allow me to coherently suppose that I, Dōna Warren, am an arrogant and irascible male physics professor at an Ivy League university.  Second, on pain of possibly collapsing their account of psychological explanation into a version of the theory theory, advocates of this conception of simulation need to justify the analogical inference without appeal to psychological generalizations.  Gordon argues, however, that this introspective/analogical model of simulation is unnecessary.  We may abandon the analogical inference by appreciating the difference between simulating ourselves in the subject’s situation and simulating the subject in the subject’s situation.  The former process would properly require an inference to the effect that the subject will think (or feel, or act) as we ourselves would think (or feel, or act) under the imagined circumstances.  The latter process, in contrast, would involve what Gordon describes as ‘a recentering of [our] egocentric map[s]’ in which we imaginatively take on as many of the subject’s character traits as possible and (consequently?) in which our first-person pronouns take the subject as their referent (Gordon, 1995, p. 55). What we have, to use Gordon’s terminology, is not a transfer of mental states from ourselves to the subject through an analogical inference, but a transformation of ourselves into the subject through an imaginative identification.  Thus, when I think, in the context of my simulation of Professor Jones, ‘I have been unfairly attacked by inferior scholars’ or ‘The editor has a vendetta against me,’ the ‘I’ and the ‘me’ refer to the subject of the simulation, Professor Jones, rather than to my actual self, and I am consequently relieved of the need for an extrapolation from myself to him. Ascriptions to Jones are underwritten, not by appending an analogical inference to a simulation of myself, but by the pronoun-replacement which occurs upon ‘stepping outside’ of the simulation.  That is to say, I don’t ascribe to Jones the belief that he has been attacked by inferior scholars by first ascribing to myself, within the simulation, the belief I have been attacked by inferior scholars and then, outside of the simulation, by assuming that Jones is relevantly like me; instead, within the simulation I ascribe to myself (as Jones) the belief that I have been attacked by inferior scholars, and then, outside of the simulation, I express the same fact by replacing my first-person pronoun with the proper name of its referent within the context of the simulation, i.e. ‘Jones.’

                None of this, however, lessens the appeal of the introspectionist component of simulation.  Quite the contrary in fact, because it assumes that I, when simulating Jones, am able to ascribe to myself (as Jones) the belief that I have been attacked by inferior scholars. How am I to engage in such self-ascription if not through a process of introspection?  Gordon responds to this worry by arguing that the reliability of our self-ascriptions can be explained without the assumption of introspective access.  Children can be trained to preface all of their assertions with ‘I believe that,’ thereby enabling them to make reliable self-ascriptions of belief long before they are able to recognize that their beliefs may be at variance with the facts.  This being the case, children can make reliable self-ascriptions of belief before they possess a full-fledged concept of belief and, ipso facto, before they can undertake an introspective survey of their beliefs: lacking a concept of belief, such children can hardly justify their self-ascriptions by asking themselves ‘Do I believe this proposition?’ Further, maintains Gordon, we don’t justify self-ascriptions of belief introspectively even after we have fully mastered the concept of belief.  We instead employ what Gordon calls ‘an assent routine,’ answering the question ‘Do you believe that the Hale-Bopp comet was larger than Haley’s comet?’ by asking ourselves the question ‘Was the Hale-Bopp comet larger than Haley’s comet?’ and by answering the former question affirmatively if and only if we answer the latter question affirmatively (Gordon, 1995, pp. 59-60).

                The important point about the assent routine, as Gordon quite correctly notes, is that it can be employed within the context of a simulation in order to generate ascriptions to others, because once we have imaginatively transformed ourselves into the other, our first-person pronouns take the subject as their referent (Gordon, 1995, p. 60). Suppose, for instance, we want to know whether or not Professor Jones believes that he has been unfairly attacked by inferior scholars.  Imaginatively identifying ourselves with Jones, we ask ourselves ‘Have I been unfairly attacked by inferior scholars?’ Noting that the answer to this question is ‘Yes,’ we employ the assent routine and ascribe to ourselves (as Jones) the belief that we have been unfairly attacked by inferior scholars.  Upon stepping outside the simulation and engaging in pronoun-replacement we ascribe to Jones the belief that he has been attacked by inferior scholars, and all of this without introspection or the deployment of any refined concept of belief.

                Of course, the assent routine will need to be supplemented with an assumption of the relevant translation when we are ascribing beliefs to individuals with whom we don’t share a common language.  Suppose, for instance, that Jones speaks only Finnish.  It would hardly do to simulate Jones and ask ourselves (as Jones) ‘Have I been unfairly attacked?’ because Jones would have no response to that question posed in English.  We must, instead, hypothesize a translation of the question into Finnish to accompany our hypothesis of only understanding Finnish, and imagine ourselves (as Jones) answering that question.  Once the object-level question has been answered within the simulation, we would ascribe the belief to Jones by supplementing our pronoun-replacement by a translation back into English.  More effectively and to the same effect, we might just imagine that the English question is the question in Finnish, and then return to the assumption of our ‘home language’ when we ascribe the belief to Jones.

                Refined in this way, the proper deployment of an assent-routine within the context of a simulation results in the ascription of externalistically individuated beliefs.  Imagine, for instance, that we want to know whether or not Oscar 2 believes that twater is good at quenching thirst.  Within the context of a simulation of Oscar 2, we ask ourselves the question ‘Does twater quench thirst?’ or rather, we ask ourselves the question translated into Oscar 2’s language, e.g. ‘Does water quench thirst?’  Noting that the answer to this question is affirmative, we deploy an assent-routine, replace our first-person-pronouns with reference to Oscar 2, translate back into our own language, and ascribe to Oscar 2 the belief that twater quenches thirst.  Now imagine that we want to know whether or not Oscar 2 believes that water quenches thirst.  Attempting to employ the same procedure, we would simulate Oscar 2, and ask ourselves the translation of the question ‘Does water quench thirst?’  Here, however, there is no such translation because there is no water on Twin Earth; the term ‘water’ doesn’t translate into the language spoken there, and Oscar 2 simply doesn’t have the concept of water.  Without this translation, we (as Oscar 2) could not respond affirmatively to the question and so, deploying the assent routine, we (as Oscar 2) would not believe that water quenches thirst.  Oscar 2, although believing that twater quenches thirst, does not believe that water quenches thirst.  Similarly, this procedure would allow us to ascribe the belief that water quenches thirst to Oscar 1, but would not allow us to ascribe to Oscar 1 the belief that twater quenches thirst.[8]  Externalistically speaking, things are as they should be.

                In summary, the only understanding of belief which is required for simulation to account for the dependence of belief on environmental factors is that thin understanding given by the assent-routine modified to incorporate translation.  Simulation is therefore able to accommodate externalism regardless of how the full-fledged mentalistic concepts are mastered.  The answer to our guiding question, ‘Can someone employing simulation generate externalistic belief ascriptions without employing strands of a theory about the mind?’ is ‘Yes.’  The simulation alternative is consequently perfectly compatible with psychological externalism.

 

3.3 Simulation and Causality

 

                It remains to be established that the simulation alternative can accommodate causal explanation in psychology.[9]  Martin Davies apparently assumes that explanations voiced via simulation do in fact get at the cause of an individual’s behavior, writing that

 

Human beings are able to predict and explain each others actions by using the resource of their own mind to simulate the psychological etiology of the actions of the others.

(Davies, 1992, p.2; my emphasis)

 

In order to see if this promissory note can be cashed out, we must recall that according to the simulation alternative we use our own mental processes to simulate the mental processes of the individual whom we are attempting to explain.  Very roughly put, we feed imagined facsimiles of the person’s mental states and environmental stimuli as inputs into our own mental machinery and we report the output.  If that output is an intention to engage in a certain behavior, then in the absence of obvious contravening factors, we predict that the individual will act accordingly. 

                Of course, we are usually interested in more than brute prediction.  We often want to explain behavior, to assist one another in understanding it, and in order to give such explanations we must be able to report salient aspects of the mental processing leading up to the simulated decision to act.  In so explaining the behavior, we are not situating the individual’s action within a system of law-like generalizations (which is not to say that our explanation could not be (mis)interpreted in this way). Rather, we are articulating our simulated decision-making process, sketching how and why the individual probably made the choice to act as she did. 

                Thus, if simulation is to provide us with genuine and enlightening psychological explanations then our simulated mental processes must be both accessible and understandable to us, at least on the cognitive level relevant to forming intentions to act, and so on.  Of course this is compatible with the psychological explanations advanced via simulation not being causal in character.  This would be this case, for instance, if our own decision-making processes, and the decision-making processes which we simulate and articulate when giving psychological explanations, fail to cause the behavior which we use them to explain.  In such circumstances, although we would be able to simulate the decision-making process behind an intention to act, and although we would be able to use such simulation to predict a behavior, and although we would be able to recount that simulated decision-making process when explaining the behavior, in no way would we be predicting or explaining the behavior by citing its cause.  In no way would we be offering causal psychological explanations.  In order to offer causal explanations, we would need what Davies seems to assume we have: access to the psychological etiology of our own behavior, and access to the simulated psychological etiology of other people’s actions.  It is this etiology which we would need to simulate in order to predict a behavior, and it is this etiology which we would need to articulate in order to explain that behavior.  Thus, if we can establish that the decision-making processes employed in simulation just are the action’s etiology, then we will have argued that psychological explanations advanced within the framework of the simulation alternative can be causal after all.

                We may uncontroversially grant that in many cases, and in most if not all of the cases of interest to us, it does seem very much as though our decision-making processes cause the behaviors which we explain by citing them, but in order to defend this feeling of causal complacency, we must take a brief detour through an objection to the simulation alternative advanced by Paul Churchland. Churchland (1989) has argued that the simulation alternative is not sufficient to provide causal explanations in part because it must ultimately rely upon some psychological theory when isolating the cause of the event to be explained.  Churchland begins by observing that even if simulation can be used as an effective predictive device, it cannot be used to explain an individual’s behavior unless one can explain the mental processes of the individual doing the simulating.  For instance, to return Gordon’s analogy, we may predict a plane’s behavior under certain circumstances by seeing how a model plane behaves under those (or analogous) circumstances.  In order to explain the real plane’s behavior, however, we must explain the behavior of the model plane; we must determine the cause of the model plane’s behavior, and in order to do that it seems we must appeal to the theory of aerodynamics, or to some other system of law-like generalizations.  Similarly, if one uses one’s own mental processes to simulate or ‘model’ another individual, and if one is to use simulation as an explanatory method as well as a predictive device, then one must be able to explain one’s own behavior.   And in order to give a causal explanation of one’s own behavior, Churchland maintains, one must appeal to some folk-psychology or system of law-like psychological generalizations; one must situate one’s own behavior and mental processes within an explanatory framework comprised of general patterns relating mental and behavioral events.  Thus, according to Churchland, the simulation alternative must ultimately rely upon a psychological theory if it is to genuinely explain an individual’s behavior.

                Gordon (1992) criticizes Churchland’s argument by noting that Churchland neglects one of the benefits of having a manipulable model, such as a sophisticated model airplane.  Specifically, such models can be used to investigate the effects of various counterfactual situations and this, in turn (employing Mill’s method of difference) can be used to isolate the cause of the event to be explained.  If, for instance, we wanted to determine why a plane stalled, we could replicate the stall-conditions with our model airplane in a wind tunnel.  We could then tinker around a bit, changing the wing design, or angle of incidence, or throttle, until eventually our model plane fails to stall.  If, say, an alteration in the wing design kept our model plane from stalling, then we may assert that the model plane’s wing design was responsible for its stalling in the replicated circumstances.  Assuming that we may extrapolate from the model, we may also assert that the real plane’s stall was caused by its wing design.

                Thus, contra Churchland, a manipulable model may allow us to determine the cause of an event without appeal to theory.  The analogy with simulation is immediate.  If, in the first person case, we want to know whether one of our beliefs caused one of our actions, we envision a counterfactual situation in which we don’t have the belief and we determine what we would do, or what intentions to act we would form, under those circumstances; if we wouldn’t engage in the action, then we conclude that the belief under consideration at least partially caused that action.  Similarly, if we want to know whether someone else’s belief caused that person’s action, we simulate the individual whose action is to be explained and within the context of that simulation we envision a situation in which we don’t have the belief; if it turns out that we (as the simulated individual) wouldn’t engage in the action unless we had that belief, then we can conclude that the belief at least partially caused the action.

                For example, let’s return once more to the wounded ego of Professor Jones and ask ourselves what caused Professor Jones to remove the journal from the departmental library.  Simulating Professor Jones, we envision a situation in which we (as Jones) don’t  believe that the journal contains a highly critical review of our paper.  Noting that we wouldn’t formulate the intention to abscond with the journal under those circumstances, we conclude that our belief that that the journal contains the critical review was a partial cause of our taking the journal.  Stepping outside of the simulation, we conclude that Jones’ belief that the journal contains a highly critical review of his paper was causally relevant to his thievery.  Of course, this is only a partial cause.  Simulating Jones again, we would note that we wouldn’t formulate the intention to remove the journal from the departmental library if we lacked the belief that this review would undermine our reputation among our colleagues, and so we may conclude that Jones’ belief that his reputation would be undermined was an additional causal factor.  Not all of Jones’ mental states would be causally relevant, however.  Upon simulating Jones we would probably find that we would decide to take the journal even if we lacked the belief that we taught on Tuesdays, or that our mother’s maiden name was Smith, and so we would be saved from attributing causal efficacy to these beliefs. The simulation approach therefore permits us to advance causal explanations by isolating which mental event, or set of events, caused the behavior to be explained.

 

3.4 Simulation and Reconciliation

 

                We are now in a position to see how the simulation alternative may reconcile externalism with causality.  Suppose, for instance, that we want to give a causal explanation for one of Oscar 2’s actions - say bringing Mary 2 twater.  We simulate Oscar 2, putting ourselves as far as possible into his situation, and following the assent-routine previously described, we ascribe various mental states to Oscar 2.  Among these mental states are the belief that twater quenches thirst, the belief that Mary 2 is thirsty, the desire to help Mary 2 quench her thirst and so on.  We then engage in a sub-simulation of Oscar 2, imagining away one of these mental states; we pretend that we (as Oscar 2) don’t believe that twater quenches thirst and we allow our cognitive mechanisms to operate upon this revised doxastic set.   If, as is probable, we note that under such circumstances we (as Oscar 2) would not bring Mary 2 some twater, then we conclude that the belief that twater quenches thirst was causally relevant to our simulated decision to bring Mary 2 some twater.  Stepping outside of the simulation, we would conclude that Oscar 2’s belief that twater quenches thirst was, in fact, a partial cause of his behavior.  Similarly, suppose that we want to advance a causal explanation for Oscar 1’s bringing water to Mary 1.  Simulating Oscar 1, we employ the assent-routine and ascribe to Oscar 1 the belief that water quenches thirst and the belief that Mary 1 is thirsty.  Simulating Oscar 1 once more, we imagine away one of these mental states, pretending that we (as Oscar 1) don’t believe that water quenches thirst, and we allow our cognitive mechanisms to run off-line.  It’s likely that this sub-simulation would fail to generate an intention to bring water to Mary 1, and so we would conclude that our simulated belief that water quenches thirst was causally relevant to our simulated intention to act.  Stepping outside of the simulation, we would conclude that Oscar 1’s belief that water quenches thirst was a partial cause of his behavior.

                We have used simulation in a seamless fashion, both to ascribe externalistic beliefs to Oscars 1 and 2 and to observe that these beliefs were causally relevant to their respective behaviors.  This, I maintain, is enough to conclude that the simulation alternative is able to embrace externalism and causal explanation simultaneously and so is in a position to reconcile one with the other.  I would like to take a moment, however, to consider a concern which might be raised by someone who is loathe to abandon the notion of narrow content.  This will have the dual benefits of staving off potential misunderstanding and highlighting how the apparent tension between externalism and causality may be traced to the prevalent assumption that psychological explanation is theory-based.

                Returning to our previous analogy, suppose that we use the simulation approach to explain two amazingly similar plane crashes which differ only in that the first plane, Plane 1, crashed in Colorado, whereas the second plane, Plane 2, crashed in Missouri.  As good simulationists, we attempt to ferret out the causes of these disasters by replicating  Plane 1’s pre-crash conditions with a model in a wind-tunnel and manipulating various factors until our model plane fails to crash.  Upon discovering that our model plane doesn’t crash if we remove its exposure to the equivalent of a 100 mph northwest wind, we conclude that the wind caused our model plane to crash in the wind-tunnel and that the 100 mph northwest wind to which Plane 1 was exposed caused it to crash in Colorado.  Because the crash of Pane 2 in Missouri was so similar to the crash of Plane 1 in Colorado, our simulation of Plane 2’s crash is identical to our simulation of Plane 1’s crash and so it yields the same results: Plane 2 crashed because it was exposed to a 100 mph northwest wind.  Of course, the wind which crashed Plane 1 wasn’t the same wind as the wind which crashed Plane 2; Plane 1 was downed by a wind which originated in Alberta whereas Plane 2 was downed by a wind which originated in Manitoba.  If we wanted to, we could individuate winds broadly, in way sensitive to their origin, and so distinguish between our two winds (Alberta-based-100 mph-northwest-wind vs. Manitoba-based-100 mph-northwest-wind), or we could individuate winds narrowly, on the basis of their intrinsic properties, and so type our two winds the same (100 mph-northwest-wind). We are now faced with the following question: Given that Planes 1 and 2 had similar crashes, and given that we have a choice between assigning causal relevance to the broad winds and assigning causal relevance to the narrow winds, which winds should we deem to be causally efficacious?  There is, I believe, a strong pull toward assigning causal relevance to the narrow wind. 

                But now the criticism of causally-relevant externalistic mental states is but one small step away.  On the assumption of narrow content, or psychological content which must be shared by physical duplicates, Oscar 1’s belief that water quenches thirst and Oscar 2’s belief that twater quenches thirst share the same narrow content, and hence Oscars 1 and 2 share the same narrow belief (belief individuated according narrow content).  We could call this narrow belief ‘belief-n.’  But narrow beliefs are components of broad beliefs (beliefs individuated according to broad, or environmentally-sensitive content), in much the same way that narrow winds are components of broad winds, and so when we simulate Oscar 2’s tokening of the belief that twater quenches thirst we are, ipso facto simulating Oscar 2’s tokening of belief-n.  (How could we simulate Alberta-based-100 mph-northwest-wind without thereby simulating  100 mph-northwest-wind?)  Furthermore, when we simulate Oscar 2 and imagine away the belief that twater is quenches thirst we are also imagining away belief-n.  Consequently, when we note that the intention to bring Mary 2 some twater would not be generated under a simulation absent the belief that twater quenches thirst we could maintain that the belief that twater quenches thirst or belief-n was causally relevant to Oscar 2’s behavior.  Similarly, when we simulate Oscar 1’s belief that water quenches thirst, we are simulating Oscar 1’s tokening of belief-n, and when we imagine away the belief that water quenches thirst within the context of simulating Oscar 1, we are also imagining away belief-n. Consequently, when we note that the intention to bring Mary 1 some water would not be generated under a simulation absent the belief that water quenches thirst we could maintain that the belief that water quenches thirst or belief-n was causally relevant to Oscar 1’s behavior.  We are now faced with the following question: Given that Oscars 1 and 2 engaged in the same behaviors, and given that we have a choice between assigning causal relevance to their externalistically-individuated beliefs and assigning causal relevance to their narrow beliefs, which beliefs should we deem to be the causally efficacious ones?[10]  Again, there is a strong pull toward assigning causal relevance to the narrow beliefs and so we would seem to have failed in our defense of the causal relevance of externalistically-individuated beliefs.[11]

                This is an important and interesting objection, and its first half, the part about the planes, is probably right.  In order for the conclusions about psychology to follow, though, the part about Oscars 1 and 2 must be relevantly similar to the part about the planes, and I maintain that there are good reasons to doubt that it is. 

                First, there are philosophical and pragmatic reasons to doubt that the notion of narrowness carries over from aerodynamics to psychology.  In a purely philosophical vein, although it makes sense to individuate winds according to their intrinsic properties and still, in good conscience, think of oneself as dealing with wind, there’s room to question that it makes sense to individuate mental states according to their intrinsic properties and still, in good conscience, think of oneself as dealing with psychological states.  If one removes all externalistic elements in the individuation of psychological states, one runs the real risk of ending up with a purely neurological, or otherwise physicalistic, individuation which looses all claim to provide us with psychological, as opposed to neurological states and so which sacrifices its relevance to psychological explanation (see Owens, 1987).  Consequently, although there is a narrow wind component to broad wind, there may not be a narrow psychological component to broad mental states.  In any event, as I mentioned in section 2.1, pragmatic concerns conspire with these worries to lead me to assume that the quest for narrow beliefs will be unsuccessful.  My purpose in this paper is to argue that everyday psychology, or some more sophisticated version of everyday psychology which adopts the practice of individuating mental states according to content, can advance causal explanations of behavior.  In the unlikely event that a good account of narrow content is in our philosophical future, it could be used to generate a psychological taxonomy which types mental states according to causal powers and so may be employed in standard, theoretically-based causal explanations.  Consequently, narrow content could well be expected to save the causal relevance of content-individuated states.  In assuming that narrow content is a chimera, I am assuming the worst-case (and, I think, more realistic) scenario for any content-based psychology, and I am arguing that psychological explanations may be causal even if there is no such thing as narrow content.  It is consequently no objection to my position to argue that it fails to account for the causal relevance of narrow content.

                The second, and related, disanalogy between aerodynamic and psychological explanations lies in the models to which we could reasonably expect these explanations to conform.  If simulation were the only way in which aerodynamic explanations could be formulated and justified, I must confess to seeing no reason why we would be inclined to prefer explanation in terms of narrow wind to explanation in terms of broad wind.  Any token of the broad wind would be a token of the narrow wind as well, and simulation is concerned with establishing that that wind, that very token wind, was causally responsible for crashing the plane.  Whether that wind is described broadly, as Alberta-based-100 mph-northwest-wind, or narrowly, as 100 mph-northwest-wind, would seem to be a matter of style over substance.  Real concern with how the wind is described reflects a real concern with wind types instead of wind tokens, and I’ve argued in section 2.3 that this concern is driven by theory formation.  A theory, with the attendant generalizations and formulas, requires explanatory states taxonomized according to causal powers because only then will the theoretical generalizations be both true and comprehensive.  Since  aerodynamic explanations can be and usually are couched in theoretical terms even if simulation is employed as a heuristic device, our inclination to attribute causal relevance to narrow winds is justifiable; only winds typed narrowly will be typed according to their causal powers and only winds typed according to their causal powers will be fit to play a role in the theoretical explanations to which aerodynamics is amenable.

                It’s at least an open question, however, whether the theory theory is a correct account of explanation in psychology and so, as I’ve argued, it’s at least an open question whether individuation of mental states according to causal powers is the way to go.  (Indeed, given the fact that individuation of mental states according to causal powers would require an account of narrow content, and given the fact that such an account may never be forthcoming, it may not even be a way to go.)  Of course, Oscar 1’s belief that water quenches thirst was causally efficacious in the formulation of his decision to bring Mary 1 some water in virtue of fact that it had the property of being about the water, as opposed, say, to being about strychnine.  I hope that nothing I’ve said so far is taken to deny this.  My point is only that we don’t need to hunt for some narrower causally relevant property on the grounds that individuating Oscar 1’s belief on the basis of what it’s about fails to generate an explanatory kind which includes beliefs about twater, schmater and other beliefs which are causally identical.  Unlike explaining plane crashes, explaining Oscar 1’s behavior needn’t involve appeal to generalizations and so doesn’t require an explanatory typing that corresponds to individuation by causal powers. Our pull toward assigning explanatory relevance exclusively to narrow-winds, therefore, need not be and should not be reflected in a similar pull toward assigning explanatory relevance exclusively to narrow-beliefs.

                In conclusion, parties to the debate regarding externalism and causality have tended to accept the proposition that if psychological explanation is causal, then mental states should be individuated according to their causal powers.  Such a commitment to causal individuation is inescapable within the widely taken-for-granted theoretical model of psychological explanation, because if the putatively causal explanations in psychology are offered by appealing to some system of law-like generalizations, then we must individuate mental states by their causal powers; individuating in a way that types causally different states the same would risk including false generalizations, and individuating in a way that types causally identical states differently would risk omitting true generalizations.  Consequently, so long as we conceive of psychological explanation along the lines of theoretical explanations advanced in other domains, we will be drawn toward the individuation of mental states according to causal powers, and externalism, which systematically distinguishes between causally identical mental states, will be perceived as a threat to causality.  

                Once we have cast our lot with the simulation alternative, however, our nomologically-driven intuitions about individuation and causality begin to fade and we see the principle connecting causal explanation with individuation according to causal powers for what it is: a methodological constraint that makes sense only from the perspective of theory construction.  By loosening our commitment to the principle which provided for the initial tension between causality and externalism, and by proving itself to be simultaneously compatible with both, the simulation alternative promises to reconcile externalism with causality.  If simulation is indeed the correct account of psychological explanation, then it’s a mistake to allow our intuitions about such explanations to be influenced strongly by facts about the theory-driven explanations offered in the physical sciences, or to blindly assume that causal explanation in psychology must refer to states individuated according to their causal powers.  By fully appreciating how psychological explanations may be set apart from the explanations offered in physics, chemistry, biology and so on, we may recognize that beliefs and desires can be both externalistic and causally explanatory.

 

Dōna Warren

Department of Philosophy

University of Wisconsin - Stevens Point

 

[WORD COUNT FOR PAPER:

10266 not including footnotes; 10930 including footnotes]

 

 

References

Burge, T. 1986: Individualism and Psychology. The Philosophical Review, 95, 3 - 45.

---------- 1989: Individuation and Causation in Psychology.  Pacific Philosophical Quarterly, 70, 303 - 322. 

Churchland, P. M. 1989: Folk Psychology and the Explanation of Human Behavior. In J. E. Tomberlin (ed.), Philosophical Perspectives, Volume 3: Philosophy of Mind and Action Theory. Atascadero, CA: Ridgeview Publishing Co. Repr. in R. Bogdan (ed.), Mind and Common Sense. New York: Cambridge University Press, 1989.

Davies, M. 1992:  Introduction. Mind & Language, 7, 1 - 9.

---------- and T. Stone 1995: Introduction. In M. Davies and T. Stone (eds.), Mental Simulation. Oxford, UK: Blackwell Publishers Ltd.

Fodor, J. 1982: Cognitive Science and the Twin Earth Problem. Notre Dame Journal of Formal Logic, 23, 98 -118.

---------- 1991: A  Modal Argument for Narrow Content. The Journal of Philosophy, 88, 5 - 26.

Goldman, A. 1989: Interpretation Psychologized. Mind & Language, 4, 161 -185.

Gordon, R. 1992: The Simulation Theory: Objections and Misconceptions. Mind & Language, 7, 11 - 34.

---------- 1995: Simulation Without Introspection or Inference from Me to You. In M. Davies and T. Stone (eds.), Mental Simulation. Oxford, UK: Blackwell Publishers Ltd.

Loar, B. 1991: Social Content and Psychological Content. In D. M. Rosenthal (ed.), The Nature of Mind. New York: Oxford University Press.

Owens, J. 1993: Content, Causation, and Psychophysical Supervenience. Philosophy of Science, 60, 242 - 261.

---------- 1987:  In Defense of a Different Doppelganger. The Philosophical Review, 46, 521-554.

Putnam, H. 1975: The Meaning of Meaning. In H. Putnam, Mind, Language and Reality, Vol. 2: Philosophical Papers. New York: Cambridge University Press.

 

 

 

 

 

 

 

 

 

May 25, 1998

 

Ms. Patterson or Mr. Lee,

 

Having taken full account of the referee’s final comments, I enclose a copy of my paper “Externalism and Causality: Simulation and the Prospects for a Reconciliation,” for publication in Mind & Language. 

 

If possible, I would appreciate a formal letter of acceptance for departmental review purposes.  You may mail it to me at the following address:

 

                                                Dōna Warren

                                                Department of Philosophy

                                                Rm. 489, Collins Classroom Center

                                                The University of Wisconsin - Stevens Point

                                                Stevens Point, WI 54481-3897

 

Thank you.

 

Yours,