Overview of Neuroscience and Philosophy

Neuroscience and Philosophy is a collection of essays based on a three-hour "Author and Critics" session at the 2005 American Philosophical Association (APA) meeting in New York. The volume is organized as follows: "The Argument" includes various excerpts from Maxwell Bennett and Peter Hacker's Philosophical Foundations of Neuroscience (2003); "The Rebuttals" include new critical essays by both Daniel Dennett and John Searle; and finally "Reply to the Rebuttals" includes a detailed response by Bennett and Hacker to both Dennett and Searle. The volume also includes an Introduction and Conclusion by Daniel Robinson.

As a whole, the volume is immensely readable and accessible to non-specialists (at least most of it; parts get rather technical, but the average intelligent reader will most certainly find this a rewarding read). A comprehensive set of end notes are included which elaborate on points made in the text, provide specific citations to support claims and objections, and also point the reader to a very useful list of additional resources to investigate.

The volume is also a good introduction to some of the main debates in cognitive neuroscience and the philosophy of mind, including the notions of consciousness, qualia, intentionality, and ordinary language philosophy. Finally, I believe that this volume exhibits some of the strengths of interdisciplinary interaction. Philosophical inquiry is enhanced when combined with empirical neuroscience; and empirical neuroscience is enhanced when combined with philosophy. Of course, with interdisciplinary work there is the risk of misunderstanding or "talking past" each other. The organization of the volume as a dialectal exchange allows for such misunderstandings to be minimized, but not entirely removed.

The Argument (Bennett and Hacker)

Following the "ordinary language philosophy" of Oxford in the early 1960s, Bennett and Hacker understand that philosophical inquiry is limited to figuring out what makes sense and what is non-sense. Scientific inquiry, on the other hand, is about figuring out what is true and what is not. Thus, it is the work of philosophy to engage in keen reflection about how we use our terms. What this means is that we must strive for conceptual clarity (what the authors call "logic-chopping") before we can hope to pursue meaningful empirical research.

Bennett and Hacker believe that much of cognitive neuroscience is afflicted with linguistic nonsense. For example, they believe that much of Descartes's misguided dualism appears in modern discussions of neuroscience (replacing the immaterial soul as the seat of thinking with the brain). Bennett and Hacker believe that this is the key bit of non-sense currently plaguing cognitive neuroscience: the misapplication of intentionality to the brain or parts of the brain, or what they call the mereological fallacy. Typical examples of such a fallacy is to say that the brain reasons, decides, or wants. But, according to Bennett and Hacker's ordinary language approach (based on Wittgenstein and Ryle) only whole persons can reason, decide, or want. In short, Bennett and Hacker object that the ascription of "intentional states" (believing that, deciding that, wanting that…) are misplaced when applied to brains or parts of brains. Only whole persons (or sufficiently complex systems — such as animals) can be the proper recipients of such psychological attributions. Thus, it is not that the sentence, "his brain wants to sort through all the incoming visual information" is false; rather, such a sentence is nonsensical (it is neither true nor false, since the attribution of intentional attitudes to the brain is a misapplication). In this sense, cognitive neuroscience must rid itself of this misleading, nonsensical way of speaking in favor of an ordinary language approach where only whole persons are the subjects of such psychological verbs.

Bennett and Hacker trace this fallacious tendency to several key players in the current debate. As such, they include in their discussion cognitive scientists, philosophers, and psychologists. After documenting how these scholars repeatedly use intentional language in this erroneous, nonsensical way, they survey the possible response routes. After consider each possible route, Bennett and Hacker conclude that in fact the majority of scholars discussing cognitive neuroscience fail to use language (especially language about intentional states) in sensible ways.

Bennett and Hacker draw several conclusions from their study. First, they claim that we cannot hope to study human cognition successfully unless we first get our language right (that is, making sure our linguistic usage avoids nonsensical fallacies. This means, among other things, avoiding the attribution of wants, desires, or thoughts (i.e., any intentional state) to the brain or parts of the brain. Bennett and Hacker insist that such intentional terms are only applicable to whole humans (or organisms suitably like human beings).

They also conclude that "getting our language in good order" will assist us in interpreting our experimental results. Without clarity of language (say, avoiding the mereological fallacy), our interpretation of scientific findings will be skewed. Furthermore, when neuroscientists are formulating new research questions, new experimental designs, and pushing the limits of our knowledge about human cognition, Bennett and Hacker insist that our philosophical and conceptual work must not lead to nonsensical presuppositions. In other words, the future success of cognitive neuroscience depends on clarifying our psychological concepts and avoiding the pitfalls of nonsense.

The Rebuttals (Dennett and Searle)

In his essay, Daniel Dennett agrees that in order to have a successful cognitive neuroscience we must diagnose and reformulate our linguistic presuppositions. For example, Dennett agrees with Bennett and Hacker that much of Cartesian dualism still infects neuroscience (shifting the emphasis from an immaterial soul that thinks to the brain that thinks, or what Dennett calls "Cartesian materialism). Dennett also agrees with the idea that we should appeal to ordinary language to resolve such dilemmas. In fact, Dennett points out that this is the very methodological suggestion that lead him to promote the term "folk psychology" (he also points out that Bennett and Hacker seem to overlook the fact that many of their points were made by Dennett some 30 years prior!). Finally, Dennett also agrees with Bennett and Hacker's rejection of "qualia" as ontologically free-floating bits of consciousness.

But Dennett's agreeable nature stops there. Most of his essay is ruthlessly critical of Bennett and Hacker (focusing his attack of Hacker). For example, one of the key arguments forwarded by Bennett and Hacker is that philosophers and neuroscientists are committing the mereological fallacy (ascribing attributes of wholes to their parts), and they think that Dennett commits this fallacy too. Dennett points out that he helped to develop this very criticism against his philosophical opponents, and does not much appreciate the sloppy research Bennett and Hacker produced that misconstrued three decades of his work.

Dennett goes on to tackle just a few key objections to Bennett and Hacker. First, Dennett questions their assertion that conceptual (philosophical, linguistic) questions are distinct from empirical questions. Dennett believes that there is a clear sense in which answers to philosophical questions are more than just "makes sense" or "doesn't make sense". Dennett believes that answers to such conceptual questions can also be "right" or "wrong". As such, the distinction maintained by Bennett and Hacker does not stand up, especially when we are engaged in what Dennett calls "naïve anthropology" (the empirical study of language). So, for example, what happens when we have a disagreement about what "makes sense"? Apparently Bennett and Hacker assume that their own native linguistic capacity is all that they need to cite in order to resolve such conflicts (appealing, as it were, to some set of "grammatical rules" that will decide which usage is sensical and which is nonsensical).

However, Dennett points out that there is no set rules for proper usage, and that when we are engaged in linguistic research, we are engaged in empirical research (not mere "logic chopping"). In short, we have no reason to privilege our own usage of language as THE correct usage. After all, we may be perverse and peculiar in what we take to be sensical, or else those with whom we disagree might be speaking a different language (i.e., a case of indeterminacy of translation). In this way, following Quine's naturalism, Dennett objects to Bennett and Hacker, insisting that there is a keen interplay between philosophy and science.

Dennett also includes a number of additional objections, including the value of empirical discoveries about the brain and the nervous system. He points out that it is an empirical discovery that our brains are engaged in processes that allow us to be conscious, to think, to feel, and to act. Moreover, it is an empirical discovery that brains, or parts of our brains, engage in processes very similar to intentional states, such as guessing, tracking, sorting, and deciding. In fact, Dennett points out that this very discovery is what lead him to develop his account of the "intentional stance". For Dennett, the intentional stance is a kind of ordinary (folk psychological) language applied to complex systems. The brain is a complex system, whose operations can be captured in meaningful ways by speaking as if intentional states applied. What Bennett and Hacker take to be "conceptual blunders" are widespread among philosophers and neuroscientists, and such "blunders" have been responsible for offering sophisticated new explanation for how human cognition occurs. In short, Dennett finds it laughable that they suggest neuroscientists give up this way of speaking, even though it has been so fruitful! (Dennett points out that the intentional stance is much more widespread than Bennett and Hacker realize: computer programmers, electrical engineers, and physicists all speak in intentional terms when discussing computers looking for the printer, thermostats realizing that it is too cold, or falling objects trying to reach a resting point.) While such widespread use of intentional language has been largely metaphorical, Dennett denies that such metaphors are devastating to empirical research. To the contrary, such metaphors are responsible for developing new (and better) explanations of complex phenomena, for promoting innovating research, and for developing comprehensive revolutions in empirical science.

Dennett concludes that Bennett and Hacker's book misses the mark. They offer a single line of attack (the mereological fallacy) that completely fails under close scrutiny. The so-called conceptual confusion Bennett and Hacker discuss is productive and helpful. While Bennett and Hacker offer no positive theories or models, if taken seriously their proposal also undermines our ability to offer rich and meaningful explanations of cognitive neuroscience.

John Searle's rebuttal is delightfully written, witty, and very pointed. His remarks are well organized and document several key objections that are devastating to Bennett and Hacker. Like Dennett, Searle begins with a few (brief) remarks regarding points of agreement. Very quickly, however, we see Searle launch his attack and make short work of Bennett and Hacker.

First, on a technical note, Searle points out that what Bennett and Hacker call the "mereological fallacy" is really what Ryle called a "category mistake". But then Searle goes on to remark that the category mistake they accuse most of the neuroscience community of making is disastrous. In fact, Searle develops a very nice line of argument against Bennett and Hacker, accusing them of committing the following fallacy: in their attempt to rid cognitive neuroscience of misguided and confusion linguistic usage, Bennett and Hacker fail to distinguish criteria for ascribing mental states from criteria for actually having mental states. Arguing that brains are the inappropriate subjects of mental states does not lead to the view that brains are in now way the locus on such psychological processes. In this sense, Bennett and Hacker are unable to distinguish the idea that while brains are not themselves literally deciding things (like deciding what to have for lunch, or which rose is white and which rose is red), the brain most certainly plays a role in such decisions. While the brain may not be the appropriate subject for such psychological verbs, it is misleading to think that the brain plays no role at all in psychology. Searle writes: "The fallacy, in short, is one of confusing the rules for using the words with the ontology. Just as old-time behaviorism confused the evidence for mental states with the ontology of the mental states, so this [strategy used by Bennett and Hacker] construes the grounds for making the attribution with the fact that is attributed. It is a fallacy to say that the conditions for success operation of the language game are conditions for the existence of the phenomena in question [i.e., consciousness]" p. 105).

Searle wraps up his critical essay with a few remarks about the appropriate relationship between philosophy and science. For example, he notes that while most philosophical problems are not solved by empirical research, there are at least some that are. Bennett and Hacker assume that such a distinction is possible, and valuable. However, Searle admits that he is unable to make a "really sharp, precise distinction… between empirical questions and conceptual questions" (p. 123). In this way, Searle aligns himself with Dennett and Quine and other naturalists who find there to be a continuity between science and philosophy.

Reply to the Rebuttals (Bennett and Hacker)

In their reply to Dennett and Searle, Bennett and Hacker maintain their position that conceptual (philosophical) questions and issues are completely distinct from empirical questions and issues. They offer a few examples, but do not offer a convincing reply to Dennett or Searle. They simply re-state their original assertion with new examples.

Bennett and Hacker also maintain their original use of the term "mereological" and insist that their original charge stands (that most of the philosophical and neuroscientific community is engaged in nonsensical, fallacious ascriptions of psychological states to brains and/or brain processes). Bennett and Hacker maintain that it is an error — a conceptual error, not a factual error — to ascribe psychological states (intentional states, consciousness, etc.) to the brain or parts of the brain. Despite Dennett's objections, Bennett and Hacker maintain that it just doesn't make any sense to ascribe psychological states to anything other than whole human persons.

Earlier in the volume, Searle objects to this suggestion by pointing out that Bennett and Hacker do not provide a positive account of personhood. What, exactly, do they mean by a person? Despite providing no such account in their original work, Bennett and Hacker suggest that talking of persons and attributing psychological states to persons makes perfect sense. While they appeal to Locke's notion of "person" as a forensic term (p. 134), they do not elaborate with many more details. Instead, what they provide is a re-hashing of Searle's extremely clear summary of Bennett and Hacker's reliance on Wittgenstein (oddly enough, Searle's explanation of what they are doing is far more coherent than the explanation provided by Bennett and Hacker!). What Bennett and Hacker's approach amounts to is that a "person" is, roughly speaking, a human being whose bodily actions are observable. Thus, we can say that a person is thinking X when they display various behaviors. Because the brain does not display behaviors that allow us to ascribe "believes that X" we cannot ascribe beliefs to the brain. This, of course, is roundly criticized by Searle (the difference between evidence for ascribing psychological states vs. actually having such psychological states), but Bennett and Hacker fail to address the main thrust of Searle's objection.

However, Bennett and Hacker agree with Searle that there is a distinction between (a) what criteria we should use to ascribe psychological states (and that these criteria are behavioral), and (b) what neurological processes or pathways or regions are involved in various psychological states (i.e., the "locus" of activity within the brain when a person is thinking). While they acknowledge this distinction, they insist that (a) points to whole persons and their observable behaviors as the proper subjects of conscious states, but (b) points to the idea that brains are merely implicated in a person's thinking (that brains are not thinking; and it's not that brains fail to think; they just aren't the right kinds of things to which thinking or not thinking apply).

At this point, the reader might begin to doubt that these scholars are really talking to each other, as opposed to talking past each other. There seems to be a lot of repeating (especially on the part of Bennett and Hacker) which fails to engage the other scholars. More on this below.

Next, Bennett and Hacker turn to Dennett's intentional stance, attempting to turn the tables on Dennett. In this criticism, Dennett suggested that his intentional stance enabled neuroscientific researchers to push their understanding further, and to provide meaningful and testable explanations. He also claimed that Bennett and Hacker's proposal would cripple neuroscientific research. They write, "No well-confirmed empirical theory in neuroscience has emerged from Dennett's explanations, for ascribing 'sort of psychological properties' to parts of the brain does not explain anything… Not only does it not explain, it generates further incoherence" (pp. 140-141). Who is right? Again, we shall return to that below.

Finally, Bennett and Hacker round out their reply by re-articulating their behavioral orientation, for it seems that they want to affirm the existence of "persons" as subjects of psychological properties, and rely on (whole) bodily behavior as the "data" for ascribing such properties, while eliminating everything else (qualia, private (non-public) mental states, etc.) except plainly-described neurological events (stripped, of course, of any talk of intentionality). How do the propose to support this view? Well, their support is rather laughable. Earlier, Searle draws a parallel with cognition by considering digestion (another physiological process studied by science). Bennett and Hacker take Searle's idea and maintain that "his stomach is digesting food" makes perfect sense (attributing the verb "is digesting" to a person's stomach is appropriate), but that it is inappropriate to say "his brain is thinking". Why? Because we can open up a person's stomach see the food being digested. Presumably, if we open up a person's skull, we won't see thinking happening. Rather, we see thinking happening, on Bennett and Hacker's account, by observing whole persons (specifically, we observe their bodily actions).

Now, this seems a little odd to use the "let's cut it open and look with our naked eye" methodology as a way of solving a conceptual debate. As a line of support for the conceptual clarity of ascribing digestion to the stomach and denying the conceptual coherence of ascribing thinking to the brain, this is not an impressive move.

It is true that when a person is thinking, we would not be able to open her skull and view her brain engaged in "thinking". But this misses the whole point: the aim of empirical cognitive neuroscience is to give us answers to exactly this sort of question (what happens in one's body when one is engaged in thinking?). Before it was discovered that the heart pumps blood, the very idea would have been dismissed by Bennett and Hacker as a bit of conceptual incoherence. The heart, after all, is not involved with circulating blood (only the whole human body does that). And besides, everyone already knows that the heart is the seat of emotions, love, sympathy, and sadness. In ordinary language, when we are suffering an emotional trauma, we call it "heart ache".

Just as it would have seemed preposterous to suggest that the heart pumps blood, it now seems preposterous to Bennett and Hacker that the brain could be conscious or could be engaged in thinking. But empirical research sometimes reveals that our linguistic presuppositions are limiting our understanding. The whole point of empirical cognitive neuroscience is to help us to understand processes that may be hard to see with the naked eye. Some complex processes are distributed across space (that is, across the span of the brain or across the span of the nervous system). In much the same way as the operation of the immune system is distributed across space, cognition is not easily viewed with the naked eye. It is up to science to help reveal parts of the world to us, and for us to respond by developing new and innovative ways of discussing the phenomena discovered.

I take it that this is precisely Dennett's view: sometimes empirical research enables us to use new linguistic conventions to capture our new understanding of the world. Bennett and Hacker seem to think that science must rely on conceptual analysis to sort out all the details in advance. Dennett's naturalism maintains that there is a reciprocity between the empirical and the conceptual. Sometimes that involves adopting the intentional stance with respect to neurological processes. Yet, Bennett and Hacker would insist that such a move undermines our ability to engage in meaningful cognitive neuroscientific research.

Thus, on one side, we have Bennett and Hacker accusing many (most?) philosophers and neuroscientists of committing what they call the mereological fallacy. They maintain that most people are engaged a fallacious ascription of psychological properties to parts of human persons (and that this undermines neuroscientific research). They believe that metaphorical usage corrupts and confuses our understanding. Their aim is to "peel away conceptual confusion" and to clarify conceptual presuppositions (p. 162).

On the other hand, we have philosophers like Searle and Dennett arguing that Bennett and Hacker's proposal would actually restrict us from speaking in otherwise very ordinary ways (including the ascription of psychological properties to parts of human persons, such as their brains). Searle and Dennett (while fierce critics of each other's work) agree that Bennett and Hacker's view would undermine neuroscientific research by placing unnecessary restrictions on the use of the intentional stance.

While this volume of essays does not provide a final answer to this riddle, important aspects of the disagreement are elucidated clearly. That is perhaps all a single volume could hope for: to better frame the debate, to clarify points of agreement and disagreement, and to lay the groundwork for further discussion. This volume does all of this in a very readable and accessible way.

© 2007 James Sage

James Sage, Ph.D., Department of Philosophy, University of Wisconsin-Stevens Point

Categories: Philosophical, Psychology