15 Feb 2013

Menary’s ‘Introduction [to The Extended Mind]: The Extended Mind in Focus’

summary by
Corry Shores
Search Blog Here. Index-tags are found on the bottom of the left column.]
[Central Entry Directory]

[Posthumanism Entry Directory]

Richard Menary

Introduction [to The Extended Mind]:

The Extended Mind in Focus

Brief Summary:

Andy Clark and David Chalmers’ article “Extended Mind” hypothesizes that our minds as cognitive systems extend into external components coupled to our mental processes. It is an externalist functionalist theory that makes use of a concept of mutual causality. Some have found flaws with the theory, but in all the criticisms are not strong enough to diminish the extended mind hypothesis.


Menary will give an overview of

Clark’s and Chalmer’s seminal article on extend mind and of the positions taken in the rest of this book.

1. Active Externalism

Clark and Chalmers (C&C) propose active externalism in response to the question ‘Where does the mind stop and the rest of the world begin?’ Active externalism is not like traditional externalism (of for example Putnam and Burge), because it “concerns the active role of the environment in driving cognitive processes.” (1) There are two interpretations of this, but C&C intend just one of them. [1] The trivial interpretation: “causally active | features of the environment influence cognitive processing in the brain” (1-2). [2] The robust reading, and the one C&C take: “some cognitive processing is constituted by active features of the environment.” (2)

Humans and external entities form coupled systems that are cognitive systems. “The coupled system constitutes the a cognitive system. It is not simply that the external features, to which the organism is interactively linked, have a causal influence on the cognitive processing of the organism; rather, the interactive link is the cognitive processing.” (2b, emphasis mine)

We must distinguish active and passive forms of externalism. [First consider Putnam’s Twin Earth thought experiment. There is a twin to our earth where all things are the same except for water. Twin water has all the same properties as our water, except it is has a different chemical composition; it is XYZ rather than H2O. Someone hundreds of years ago would not know chemical compositions. On Earth one’s concept of water refers to one thing, and on Twin Earth it refers to something else. Nonetheless, the mental states of the people thinking ‘water’ on either planet is the same. So internal factors alone are not enough to explain the semantic reference to water. Rather, external factors, the different compositions of the waters, help determine the semantic reference relation involved in thinking water. The external features are ‘distal’ because they are outside our body, and they are historical, because there was a causal chain of events (in our interacting with the world) that led to the particular reference relation.] C&C note that with regard to Putnam’s Twin Earth thought experiment, if we teleport to Twin Earth, we still are thinking H2O and not XYZ, because of the historical chain causing us to have that reference. Thus the external factors are not interactive in the present. This is then a passive sort of externalism. C&C’s externalism, however the external factors co-constituting our cognition have a synchronic effect on us, so it is an active externalism. (2-3)

2. Causal Coupling

Humans and external entities form coupled cognitive systems when the human organism and the external entity are linked in a two-way interaction. There are criteria for this cognitive coupling.

[1] All the system’s components play active causal roles.

[2] Jointly the system’s components govern behavior in the same way that uncoupled cognition normally does.

[3] Removal of the external component causes the system’s behavior competence to drop, just like when removing a part of the brain.

[4] Thus, such coupled processes are no less cognitive than normal uncoupled cognitive processes, even though they may not be entirely in the head. (3)

For C&C, external features are just as causally relevant as our brain’s internal features are. We will examine two interpretations of this causal coupling. [1] Asymmetric influence [external features can causally influence our cognition without being a part of it]: “environmental features have a causal influence over inner processes. It may still be the case that we can change the external environment and that affects competence and behavior of the subject.” (3d) If someone takes our diary, that eliminates a cause for our recollections, but it does not remove a part of our memory. (3d) [2] Symmetric influence [internal and external features causally interact in a mutual way]: “the inner and outer features have a mutually constraining causal influence on one another that unfolds over time.” (4a) So the diary not merely causes us to recall something. There is the external process of checking the diary and the internal process of recollection in our brain. Those two processes are concurrent and they “jointly govern my future behavior.” (4) Menary calls this cognitive integration.

Clark elaborates these coupled processes in his book Being There, where he explains continuous reciprocal causation, continuous mutually modulatory influences linking brain, body and world (Clark 163; Menary 4) So we consider two systems, system S and system O. There is continuous reciprocal causation between these systems when each one is both affecting the other while simultaneously also being affected by the other. We already  know that parts of the brain work in this continuously coupled way. Clark also gives the example of the dancer. First consider the continuous reciprocal causation between the dancer’s neural-states and her body orientation. The neural states influence how the dancer poses all while those poses influence her neural states. Concurrently, this dancer’s movements both affect and are affected by her dance partner’s movements (which themselves are mutually affected by neural states). [This is like Deleuze’s notion of rhythm in Spinoza’s affection, see the end of section 6 of my paper “Body and World in Merleau-Ponty and Deleuze”:

Our active self-affection and adaptive interaction with the world around us is what Deleuze here calls "rhythm." He also offers the example of swimming through a powerful wave. When we collide with the wave, its affection begins to decompose our body. Yet, by self-affectively altering the arrangements of our own body's parts, we may swim in conjunction with the wave and together form a larger composite body. Deleuze suggests another illustration to explain more clearly how affective rhythm involves couplings of continuous affective variations. He has us consider a dual improvisation of a violin and a piano. On the one hand, each one needs to improvisationally choose its own development. Yet, the musicians' decisions will influence how the other plays in concord with it. So, in order for both instruments to maintain their differential co-composition, they must make self-modifications that are differentially compatible with those of the other player. (Shores 203)


So are such coupled systems one system or two? And also, are there not coupled systems which are not involved in cognitive behaviors?

Although we can identify the relevant components, and factorize them into internal and external components, the nature of reciprocal coupling makes it difficult to study the components as separate systems because they are continuously influencing and responding to one another. They are coordinating with one another to produce behavior. Insofar as brain, body, and world can be shown to be reciprocally coupled in this way, we can consider them to be a coupled system. However, we are still not in a position to definitively say when a coupled system is a cognitive system, because there might be coupled systems that are noncognitive. (4d emphasis mine)

The parity principle is supposed to help make this distinction.

3. The Parity Principle

So we have process at work in our head that we consider cognitive. We also have processes in the world. Many of those processes function to accomplish certain things. Some might accomplish the same things that processes in our head accomplish. So although these systems are not limited to confines of the head, because they accomplish what cognitive processes in the head accomplish, we likewise call them cognitive processes. Thus the two processes have parity with each other. Hence we see the functionalism at work in extended mind theory:

As long as a process has a cognitive function then it does not matter where it is located. If it plays the right sort of role and is causally integrated with other cognitive processes, then it is part of the system of processes that constitute a person’s completion of a cognitive task. (5)

[In this entry I discuss Putnam’s functionism. I quote from it:

For Hilary Putnam, isomorphism is a criterion for artificial intelligence. Yet, we must first expand our notion of function. Putnam still means it as an assignment relation between isomorphic systems, but the ‘systems’ in this case are not static like sets of numerals, but instead are sequences of successive states, as though one necessarily brings-forth the next. To illustrate, consider when we enter ‘2’ ‘+’ ‘2’ into an electric digital calculator and then press ‘=.’ After doing so, ‘4’ displays on its screen. Likewise, when we move two abacus-beads, then two more, the result is four beads grouped together. Here we might say that there are two sorts of functions: trans-systemic functions and intra-systemic functions. Trans-systemic functions are the ones that assign states in one system with states in another (4 displayed on the calculator screen; four abacus beads). Intra-systemic functions are ones that assign within the same system an ‘output’ state to an ‘input’ state, if you will (we assign the resulting four beads to the movement of two sets of two).

So, Putnam defines functional isomorphism in general as “a correspondence between the states of one and the states of the other that preserves functional relations.” (This correspondence between states is a trans-systemic function; the preserved functional relations are intra-systemic functions). Thus, if “state A is always followed by state B” in one system, there will be corresponding states in its isomorphic counterpart: the digitally displayed

correlates to the abacus’


because both follow unique states that themselves correspond with one another.[vi] We might also apply the more common usage for the term function, and note that these devices function equivalently – but not identically – insofar as both are capable of the same calculations.

Putnam notes that these devices are physically realized in utterly different ways; for, the calculator’s integrated circuit bears little resemblance to a rack of beads. Hence his provocative conclusion: “so a computer made of electrical components can be isomorphic to one made of cogs and wheels or to human clerks using paper and pencil;” for, humans too can begin in an equivalent state A (receiving the mathematical problem ‘2 + 2’) and result in an equivalent state B (writing the number ‘4’). Hence, if a machine, software-program, alien life-form, or any other such alternately physically-realized operation-system were functionally isomorphic to the human brain, then we may conclude, says Putnam, that it shares a mind like ours.[vii]


[vi] We know he must be speaking of both trans-systemic and intra-systemic assignment functions, because he writes, “To start with computing machine examples, if the functional relations are just sequence relations, e.g. state A is always followed by state B, then, for F to be a functional isomorphism, it must be the case that state A is followed by state B in system 1 if and only if state F(A) is followed by state F(B) in system 2. If the functional relations are, say, data or print-out relations, e.g. when print p is printed on the tape, system 1 goes into state A, these must be preserved. When print p is printed on the tape, system 2 goes into state F(A), if F is a functional isomorphism between system 1 and system 2.” Here we see that the functional relations are the “sequence relations” within one system, and the functional isomorphism F is the correlation-relation between both machines. Hilary Putnam, “Philosophy and Our Mental Life,” Mind, Language, and Reality, (Cambridge: Cambridge University Press, 1975), p.292.

[vii] Putnam, p.293.

So brain cognitive processes can be functionally isomorphic to brain-external system, despite being realized in physically different ways, thus coupled systems can be cognitive.]

Some critics take issue with the parity principle. They think that it is based on the assumption that externalized cognitive processes are so similar to strictly internal ones that we may consider them both as cognitive.

Menary thinks that critics miss the point that coupled cognitive systems are functionally similar even if drastically dissimilar in many other ways (6).

Some critics, for example, point out the biological memory is susceptible to such factors as recency, interference, and chunking, while non-biological memories stored in notebooks are not. We need to use two different explanatory approaches to explain the workings of biological and non-biological memories, thus they are not similar enough to both classify as the same sort of cognitive process.

Yet, some show how internal and external memories compliment one another.

4. Portability, Reliability, and the Linguistic Surround

Another objection has to do with the portability and reliability of coupled systems. Our biological cognitive systems remain whole even as they move from place to place, and the conjunction of its parts is reliable. However, external coupled cognitive systems do not reliably go along with us wherever we go.

But the brain can be damaged and thus its integrity is not completely reliable. It is conceivable that externalized systems can be just a reliably available to us as parts of our brain are.

In fact, C&C note, our biological systems have evolved so that they often use reliably available parts of the external world; for example, our vision is often aided by visual ‘clues’ in the external world.

Consider also the linguistic environment we are immersed in. We are coupled to it, our brains and it are reciprocally causally coupled. So we should already see the externality of mind in the lingual systems that are a part of how we think and communicate and that are reliably there for our usage.

5. From Cognition to Mind

So we see how extended mind theory deals with the way cognition can extend into the external world. C&C also think the mind extends as well. So consider if features of the external world help constitute our beliefs by playing the right role in driving our cognitive processes. In that case, C&C will argue, our minds extend into the world. (8)

The make this argument with their Otto and Inga example. Inga hears about an exhibition at the Museum of Modern Art (MOMA). She recalls from her biological memory that it is on 53rd street. Otto has Alzheimer’s, so he depends on a notebook to retrieve such information. He always carries it. He consults it, and sees that MOMA is on 53rd street. (8-9)

The notebook serves the same role as Inga’s biological memory, so they are on par.

C&C’s argument calls on their commitment to functionalism: as long as information plays the relevant role it is a belief, regardless of location. Hence, the mind extends into the world. (9)

Yet it is debatable whether "Otto and his notebook display the same causal profile as Inga and her biological memory”. (9)

6. Criticisms of the Extended Mind and Responses

Extended Beliefs

One objection is that there are two steps, two beliefs [in Inga’s case there is just one, so they are not on par]: [1] Otto believes the information is in the notebook, and [2] Otto believes the knowledge he finds in it. (9-10)

Clark thinks this objection does not work because the notebook is transparent equipment. But Clark’s response forgets this difference: Inga merely remembers the info, Otto must remember that the notebook remembers the info. (10)

John Preston objects that there is no parity, because Otto does not have ‘first-person authority’ over his beliefs: he must consult his beliefs before he knows what they are. (10)

Preston also objects that normally we do not credit our external tools with achieving the cognitive tasks they helped us accomplish. We do not give the calculator equal credit for doing math. (10-11)

However, we do sometimes need to consult external sources (like diaries, beer mats, etc.) to know what our beliefs are.

The Coupling-Constitution Fallacy (Fallacy)

The coupling-constitution fallacy, according to Adams and Aizawa, is saying that the external object or process coupled with the cognitive agent is part of the agent’s cognitive apparatus.

The fallacy is based in the distinction between causal relations and constitutive relations and “the fact that object or process X is coupled to object or process Y does not entail that X is part of Y ” (ibid., p. 68). [11]


The second line of attack is closely allied to the first: because we need to ask which processes are candidates for inclusion in the “kind” cognitive. Schematically the argument strategy runs like this: 13 neuronal (and therefore cognitive) processes have property X; nonneuronal processes do not have property X ; therefore nonneuronal properties are not cognitive. [12a]

Menary replies that the correct formulation should be:

X is the manipulation of the notebook reciprocally coupled to Y — bodily processes, including neuronal ones — which together constitute Z, the process of remembering. Once we have this picture, it is easy to see that Adams and Aizawa have distorted the aim of the extended mind. The aim is not to show that artifacts get to be part of cognition just because they are causally coupled to a preexisting cognitive agent, but to explain why X and Y are so coordinated that they together function as Z, which causes further behavior. (12)

Clark responds in a similar manner by saying that we are trying to explain how some object that is not normally thought of as cognitive becomes a part of a cognitive system. (12)

Adams and Aizawa might be committing the fallacy of composition, which is saying that the parts of a system must have the properties of the whole. (13)

Hurley and also Ross and Ladyman note the misuse of the causal-constitutive distinction. And criticisms are based on containment, meaning that properties are explained by the properties on the lower level, however wetness is an emergent property. (13)

This containment metaphor does not belong in cognitive science. (13)

Fleeting versus Persistent Cognitive Systems

Rupert notes that Otto is only intermittently coupled to his notebook, thus the system they form is fleeting, and there would be no stable and persisting individual to study. (13-14) Rupert thinks the coupling is unreliable; however, supervenience has been based on such a reliance, for example, “Mental properties are dependent on neuronal properties in an especially clear way; therefore the brain instantiates mental properties.” (14)

Just because the system’s parts are external does not mean they have an unreliability that should disqualify them as system parts. For example, a spider relies on its web for eating.

spiders are clearly dependent on their spider webs to catch prey—the spider’s prey-catching system consists of both spider and web (and spider’s webs are fleeting systems if anything is). The organismic process extends beyond the boundary of the body of the organism in this case. Similarly, the caddis fly larva collects small stones and shell fragments from the riverbed and binds them together with a kind of secreted cement (Dawkins 1982). The caddis fly larva then lives in and carries this new home around with it on the riverbed for its larval period. Humans with their linguistic surround, speech and writing, are in a similar situation. They must create and maintain delicate and intricate linguistic webs as part of their cognitive processing. [14, boldface mine]

Rupert’s problem is that cognitive systems have their capacities only for the fleeting period when the extended systems are coupled to them. (14-15)

Perhaps the response to Rubert is that we have long-standing dispositional capacities. So although Otto might lose his notebook, he is still able to make a new one. (15)

Derived and Underived Content, or The Mark of the Cognitive

Adams and Aizawa also object that external parts of the system cannot be cognitive, because cognitive states involve intrinsic, non-derived content. “They do, however, draw a fundamental distinction between vehicles with conventionally determined (derived) content and vehicles with naturalistically determined (nonderived) content”.(15d)

In response Clark notes that the overlapping parts of Venn diagrams have conventional content and yet are still part of cognitive processes (15-16)

Adams and Aizawa object that things like words and signs get their meaning through a social convention, but mental representations do not. (16)

They then “reject the derivation of the content of the image from the external Venn diagram, the social convention governing the intersection of Venn diagrams is ‘not a fact about the constitution of the content of a mental image of the intersections of [Venn diagrams]’ (ibid., p. 72).” [16]

And they note that cognitive content cannot be determined by social convention (we cannot change the meaning of neuronal states like we can do for the word “cool”). (16)

Menary notes in response to Adams and Aizawa that the image of a car does not automatically through naturalistic means give us the concept of the car, because that concept requires such conventions as what the car is used for an so forth. (17)

Clark responds to them by reminding that it is not a matter of whether Otto’s notebook is cognitive but rather if it and Otto form a cognitive system. So it does not matter if the notebook lacks intrinsic contents, because the notebook itself is not a cognitive system. (17)

Proponents of extended mind should provide empirical examples rather than imaginary ones. (18)

Scientific Kinds

Extended mind uses causal processes, but psychology has not shown lawlike regularities of intercranial processes, only intracranial ones. (18)

The processes involving externalized memory are so different from purely internal ones. (18)

Clark, Sutton, and Menary think these differences do not matter provided that the external and internal processes are sufficiently complementary and integrated. (18) But critics might demand for scientific accounts of the lawlike regularity of complementary and integrated systems. (18)

Clark also suggests that  “cognitive is as cognitive does,” so we should not say something is cognitive on account of its causes but rather on the basis of its effects.

Hence, we should not look for a distinctly unified set of similar causal properties that give rise to causal regularities, but instead expect to see a “motley” crew of internal and external resources that produce regular effects because of a looser coordination “poised” in such a way that characteristically cognitive behavior is produced (cf. Otto and his notebook). (19a, emphases mine)

Hurley provides a taxonomy of externalisms, and distinguishes “what” and “how” varieties.

“What” explanations explain mental states in terms of their personal-level content types or phenomenal quality types. “How” explanations explain the workings of the processes and mechanisms that enable mental states (that are of a content or quality type). “What” versions of externalism are familiar as the standard content externalism of Putnam and Burge, although less so in the “what” phenomenal sense. However, “how” externalism is the newer and more radical version of externalism, in that it is committed to enabling mechanisms, processes, and vehicles being external. In one obvious sense the extended mind falls within the “how” externalist camp, because it identifies external processes and vehicles as enabling cognitive processes and mental states. (19)

Hurley further distinguishes cultural and noncultural cases of extension.

In the cultural cases, an external artifact enables mental states or cognitive processes (Otto’s notebook being the test case for C&C), whereas in noncultural cases extended sensorimotor dynamics extend enabling processes. This goes some way to answering Adams and Aizawa’s charge that the extended mind creates an unscientific motley. (19)

Then “The remaining essays in the volume provide a variety of ways in which the extended mind project can be pursued.” (19)

7. The Second Wave of Extended Mind Arguments


The articles toward the end of the book look for new directions to take extended mind. (19-20)

Wilson thinks we need to look at essences rather than intentionality to see if activities are cognitive. (20)

Sutton distinguishes first and second wave arguments for extended mind. The first wave is based on the parity principle, where external processes function in the same manner as cognitive processes in the head. Second wave arguments are based on the complementary principle. They say that external processes and vehicles can be radically unlike external ones. (20)

Menary (in his article later in the book) says that there are

two ways in which extension, or integration, can happen: one involves integration through sensorimotor activity, the other through the manipulation of external representations. What Menary adds to the discussion is the importance of normativity for any account of extension/integration, whether it is primarily biological or biocultural. (21)

Wheeler says that extended mind is an extended functionalism.

It is a thesis that takes the bodily manipulation of external vehicles as constitutive of cognitive processes. Wheeler argues that this commits us to a functionalist account of cognition, where cognitive processes and vehicles are multiply realizable, insofar as the stuff in which the processes are realized allows for the function to be discharged. Wheeler argues that such multiple realization of functions is often found in nature, in which case, contra Adams and Aizawa, extended functionalist minds may turn out to be actual. He then goes on to argue that certain objections to the extended mind can be dealt with by providing a high-level liberal grain of functional analysis—such as that raised by Rupert (2004) and Sprevak (forthcoming). [21]

Thus Wheeler focuses on the functionalist aspect of extended mind. (21-22)

Rowlands argues for an extended account of intentional conscious states. (22)

For Rowlands, intentional acts disclose or reveal the world.

this form of disclosing activity does not supervene exclusively on what is inside the head. Rowlands goes on to argue that it follows from this that consciousness is extended into the world via disclosing and revealing activities that are intentional acts. (22)

Wilson’s, Menary’s, and Sutton’s focus on cognition as an activity goes with Rowland’s extended account of intentional consciousness. (22)

Cowley and Spurrett look at how language develops on account of extended mentality. (22-23)

Cowley and Spurrett give the example of infant-caregiver dyads to illustrate extended cognition at work in language. (23)

Menary concludes that this book contains essays in support and in critique of extended mind, contributing to a debate that will continue for many years to come. (23)

Menary, Richard. “Introduction: The Extended Mind in Focus” in (ed.) Richard Menary The Extended Mind, Cambridge/London: MIT Press, 2010.

Shores, Corry. “Body and World in Merleau-Ponty and Deleuze” in Sudia Phaenomenologica, vol.12, 2012, pp.181-209.



No comments:

Post a Comment