Phil 860 – Topics Phil
Science Professor John Symons
Paper # 1 – Due Sep 14
The Extended Mind Hypothesis:
Objection and Defense
Jenny Lorraine Nielsen
Introduction
In “The Extended Mind”, Andy Clark and David Chalmers (1998) present a claim that mind extends beyond brain and body into the physical world via coupling of organism and external resources in order to hold mental states (such as beliefs) “constituted partly by features of the environment” (12, 18). One strong criticism of the hypothesis comes from Ken Aizawa and Fred Adams (2001), who argue in “Defending the Bounds of Cognition” that Chalmers and Clark commit a fallacy by confusing the causal relationship between the brain and external objects with a constitutive relationship (2). They further maintain that such objects cannot make up cognition as they lack the “nature” of “what constitutes the cognitive” (2001, 3). In the sections that follow, I will explicate this objection and then present counterarguments that proponents of the extended mind hypotheses may use to overcome it.
The Objection
Adams and Aizawa (also referred to herein as A&A) argue that Chalmers and Clark (also referred to herein as C&C) confuse causal and constitutive relationship between concepts. This forms a strong objection to the extended mind hypothesis. In order to better understand this objection, it is first helpful to briefly explain “causal” and “constitutive” relations (something A&A do not do in their argument). After exploring this, I will examine a few example thought experiments in “The Extended Mind” (notably the methods of object rotation and the issue of Otto’s notebook) via a critical lens while keeping Adams’ and Aizawa’s primary objection in mind.
Constitutive via Causal
In general, causal relations are relations consisting of ordered or otherwise organized events (e.g., “Incident A happened in circumstance C due to events 1, 2, and 3 and their spaciotemporal organization”), whereas constitutive relations are generally relations consisting of properties and organization of parts in composition (e.g., “System A is like this because of component parts 1, 2 and 3 and their properties P and organization O”); for a further discussion, see Ylikoski (2013).
A clear example of this distinction comes up if you strike a match and light a candle, causing it to burn; by performing this operation, you and the match do not become part of the constitutive properties of the candle (which we imagine include a wick immersed in wax)—you instead enter into a causal relationship with its activity, burning. While operation “striking match” may be coupled to the event “candle burning”, it is erroneous to say that this makes the match a part of the candle! An object or operation may be coupled to another object causally, but this does not make it part of that object. Along this line, Adams and Aizawa argue that external objects in the physical environment act upon cognition via cause and effect, but this does not mean that they constitute (make up a part of) cognition or allow cognition to “extend”.
Problems in Cyber Land: Methods of Rotation
Critics arguing based on “constitutive” vs “causal” relationships as described above would do well to carefully examine the three cases of human problem solving presented by Clark and Chalmers (1998) in which a human accomplishes rotation of an object by three different methods: (1) via mental rotation in the imagination, (2) by clicking a button so that a computer generates an animation of the rotation, and (3) by access of a chip embedded in their brain, which rotates the object for them (p. 7). Clark and Chalmers crucially maintain without explanation that case (3) is “clearly...on par” with case (1), and then since (2) and (3) are computationally equivalent, that we then must admit that (1) and (2) are also on par with one another. In their intuition that (1) and (3) are “clearly…on par”, Chalmers and Clark seem to imply that the location of the chip clearly makes it a part of the constitution of the brain and thus mind of the human (it’s “in the head”!), yet critically the entire point of the chip is that no mental rotation must be performed—no exercise of imagination is involved—which makes it more on par with watching the animation in (2) than the mental activity in (1). Critics easily maintain that whereas in (2) and (3) the computer activity is causally connected with cognition, there is no resulting constitutive relationship between computer and cognition (and thus no extension of mind). Aizawa and Adams might further argue that the computer lacks “nature” of cognition or the “mark of the cognitive” as they maintain that cognition is a causal process that must contain “non-derived content” (a concept beyond the scope of this paper but for discussion, see Adams and Aizawa, 2001; Aizawa, 2005).
Otto’s Notebook
We next can critically approach C&C’s discussion of Inga and Otto and their respective ways of getting around town. Inga and Otto separately navigate to the museum—Inga by memory, and a forgetful Otto by reference to a notebook containing directions to the museum. Chalmers and Clark maintain that so long as use of the notebook is substituting for events that if occurring “in the head”1 would be considered part of cognition, then that operation of using the notebook itself is part of cognition (1998, 8). But does the notebook really function as a process that would occur in the imagination? The notebook informs Otto of certain realities, but presumably Otto (or some other cognitive agent) wrote down the items in Otto’s notebook (thus performing another cognitive operation!), and the cognitive task of reading the notebook, and the notebook’s ability to cause cognitive events to happen in Otto’s imagination, is not the same thing as the notebook itself performing cognitive operations or participating constitutively in Otto’s cognition. A&A are explicit that the notebook is not cognitive! While it may prime or trigger certain cognitive events, it does not stand in for the act of cognition.
1 “In the head” equivocates mind and brain unnecessarily; it’s just as well to use “in the imagination” wherever that might take place
The Defense
Now, we come to the task of defending the extended mind hypothesis against the protests highlighted above. Recall that Adams and Aizawa argue that while external objects and physical surroundings may act upon cognition via cause and effect, they do not make up cognition, as they lack the “nature” of “what constitutes the cognitive” (2001, 3). Our first line of defense against A&A’s accusation of fallacy is to reply that their accusation itself is partly fallacious as a matter of direction. In our next several sections, we will first examine this directional fallacy, then re-examine the problem of the notebook, pick apart the barriers between causal and constitutive relations in order to defend against accusation of fallacy, turn to an example of a gun involved in crime, and finally return to the problem of rotational methods in order to clarify our defense.
The Directional Fallacy
Crucially, being cognitive in nature or acting as a cognitive agent is not the same as constituting a part of a cognitive thinking system. It is fallacious to say that B is not composed of A is logically equivalent to disproving that A is composed of B. Stating components which may constitute a concept are not themselves composed of that concept is not to say that the concept is not composed of those components. As mentioned previously, a good deal of A&A’s “Defending” is spent arguing that an object (such as the notebook or pencil discussed by Clark and Chalmers) is not itself cognitive or performing cognitive activity, in order to reason that they cannot be constitutive of cognitive activity. Yet it is spurious to maintain that objects must be constituted of components individually instantiating the system they constitute, just as it is specious to say melted wax, a cotton rope, or melted crayons must individually exhibit “candle-itive-ness” because they may constitute part of a candle.
Of Neurons and Notebooks
In a reply to Adams and Aizawa, Clark responds that a clump of neurons is no more cognitive in nature than a notebook (2001, 35); neither the notebook nor cluster of neurons is doing anything cognitive considered alone. If a clump of neurons can participate in cognitive systems without being cognitive “in nature”, why not pencils or notebooks? Critics may respond that the neuron clump is arguably an intrinsic part of the cognitive agent and that the notebook is not, but proponents may respond (a) that cognition may not be limited to the agent, and/or (b) that the notebook has become “part of the agent’s cognitive routines” (Clark, 2001, 5). Both of these responses are constitutive claims, and so we have come to a point we must defend against the “causal-consitutive” fallacy C&C have been accused of making.
Breaking the Constitutive-Causal Barrier
Here we return to the accusation that C&C fallaciously confuse causal and constitutive relationships by suggesting causally impacting a cognitive process constitutes a part of a cognitive process. I will first point out that in their own attempt to define cognition, A&A describe cognition as “a species of causal processing…” and thus, per definition of process, we are immediately inclined to admit that cognition is at least in part composed of causal chains of step by step events. (Indeed, a process may be composed of several sub-processes; see process “checking the mail” which typically involves the subprocess “walk to the mailbox”).
Now, if a causal relation of steps or subprocesses may constitute a process, we are forced to admit a distinct boundary between causal and constitutive relations has already broken down. (This breakdown actually happens quite frequently in sciences, as described in 2010 by Don Ross and James Ladyman.) So critics may be forced to admit cognition is constituted in part of causal chains, which challenges the concept of a “causal-constitutive fallacy.”
But now we return to the operation of striking a match which causes the process “candle burning.” Just because a causal event can be constitutive doesn’t mean it necessarily is. Simply existing in step by step processual motion with a secondary process does not make the two processes a single process or one process a constitutive part of the other. Nevertheless, it is possible to argue that objects in the extended world (such as Otto’s notebook) may partake as constitutive sub-processes of the process of cognition.
Guns and Notebooks
For an unpleasant but useful analogue, consider the concept “murder”. While the expression “murder” is associated with a heinous act of grave moral turpitude committed freely by a morally knowledgeable and cognitively capable agent a.k.a. murderer, the process of murder almost invariably contains the subprocess “use of weapon”, e.g., “firing of gun”. No rational agent would argue that a gun is independently capable of murder or that a gun used in murder is a murderer. Nevertheless, anyone would vouch that a gun may be both a causal and constitutive part of the process “murder”, and some would further argue that without access to a gun, some individuals would not have been primed by its very presence to murder in a momentary fit of extreme pathos. Now let us return to Otto’s notebook. If the notebook enables Otto to have thoughts that he would not have without it (or that he would find very difficult without it), just as the gun enables actions not possible (or much more difficult) without it, what makes the notebook less a part of his thinking process than the gun is part of a violent action?
Constituting Mind
If a cognitive task is completed with seamless integration of external inputs necessary to its completion, it becomes difficult to argue that the external inputs are not in some way part and parcel of the cognitive task, just as a gun may be a constitutive element of crime or a walk to a mailbox a constitutive element of the act of checking mail. Still, even conceding the notebook participates in a cognitive routine, a critic might argue that these extended routines or systems do not constitute “mind” or extension thereof. They could easily argue that mind is roughly equivalent to the boundaries of a cognitive agent, and thus that mind and agent is separate from tools used to improve cognitive function. Proponents of an extended mind could reply that mind itself is equivalent to an organization of cognitive processes performed by an acting agent and extended systems, or that the agent itself is extended by coupling with external objects (Clark and Chalmers, 1998, 18).
In their argument, Clark and Chalmers focus on carefully defined types of couplings involving aspects that “were it done in the head, we would have no hesitation in accepting as part of the cognitive process” (1998, 8). Extending this argument, if Otto’s beliefs are part of his identity of self, and the beliefs happen to be constituted in notebook rather than in memory, the notebook is an actual part of Otto’s self (Clark and Chalmers actually argue for this, see 1998, 14,18). Depending on the beliefs or memories the notebook contains, this line of reasoning becomes emotionally compelling; if Otto has dementia and only remembers the name of his adult son and deceased wife via the notebook, is there a way to argue that the notebook is not a key constituent part of Otto’s self or sense of self? Whereas in a reply to A&A, Clark mentions the potential for critics to fear “devaluation” of “the cognitive core” (27, 2001), it’s actually just as frightening to subtract such a critical item of extension from Otto’s identity.
I risk venturing slightly off course at this point to mention a side argument that without input from and coupling to the environment more generally, mind may not be able to exist at all. Although the mind performs operations distinct from outer events (such as dreams), it is almost impossible to imagine mind or dream completely in nihilo. Just as animals cannot live without oxygen used internally but sourced externally, without input from the external environment of some kind or another, mind or thought might cease. While dependence on the external is not equivalent to constitution by the external, an emerging view that mind is embedded in the external and vice versa becomes rationally compelling, and thus the concept of “extension in the external” less challenging.
Return to Cyber Land: Object Rotation Methods Again
Now we return to the problems brought up by Clark and Chalmers of the humans performing operations in mental and cyber space. What can we say about Method (2) at this point? If a human being pushes a button and watches an animation of an object rotating in space, is the human performing a cognitive activity simply by watching it happen and realizing what happens? If we accept that Otto’s notebook is a part of his mind or cognition, we are inclined to accept that the computer animation is part of a larger cognitive system. We observe that the animation is seamlessly filling in for a process that could be done more slowly via the imagination, but if instantiated outside the imagination still constitutes a crucial part of a cognitive operation. This interpretation puts (2) essentially on par with both (3) and (1), and we agree with C&C that the boundary between Nielsen, The Extended Mind: Objection & Defense Page 1 0
skull and world without has become irrelevant. 2
2 While this may have strange repercussions for teachers trying to vouch to students that googling answers with a smart phone is not “thinking”, this concern is beyond the scope of my paper!
Conclusion
We see that one strong criticism of the extended mind hypothesis is challenged by similar possible problems as the extended mind hypothesis itself, and that it is possible to rescue the hypothesis from the accusation of a fallacy (and point out a possible fallacy in the accusation). As a last note upon departure, I suggest that to avoid such accusations of fallacy, careful definition of “constitutive” and “causal” explanation is called for. Furthermore, a deeper inquiry into definition of the boundaries, constitution and causal relations of processes and objects (or even processes vs objects) may serve both critics and proponents of the extended mind hypothesis in sharpening and strengthening of claims.
Bibliography
Adams, Fred and Aizawa, Kenneth. Defending the Bounds of Cognition. (2010). Menary, R. The Extended Mind (Life and mind). Cambridge, Mass.: MIT Press. Available at http://fitelson.org/prosem/adams_aizawa.pdf (PDF page numbering cited in text).
Aizawa, Kenneth. The Coupling-Constitution Fallacy Revisited. Cognitive Systems Research. Volume 11, Issue 4, (2010), pp. 332-342 Available at: https://www.sciencedirect.com/science/article/abs/pii/S1389041710000343
Aizawa, Ken, and Fred Adams. "Defending Non-Derived Content." Philosophical Psychology 18, no. 6 (2005), pp. 661-69. Available at: https://www.tandfonline.com/doi/abs/10.1080/09515080500355186?journalCode=cphp20
Clark, Andy. Coupling, Constitution, the Cognitive Kind: A reply to Adams and Aizawa. (2001) DOI: 10.7551/mitpress/9780262014038.003.0005. Available at:
https://www.researchgate.net/publication/287635254_Coupling_constitution_the_cognitive_kind_A_reply_to_Adams_and_Aizawa
Clark, Andy and Chalmers, David. The Extended Mind. Analysis. 58 (1998), pp. 7-19.
Available at: https://era.ed.ac.uk/bitstream/handle/1842/1312/TheExtendedMind.pdf
Ross, Don and Ladyman, James. The Alleged Causal-Constitutive Fallacy and Mature Sciences. Menary, R. The Extended Mind (Life and mind). Cambridge, Mass.: MIT Press. Available at https://www.researchgate.net/publication/322487079_The_alleged_coupling-constitution_fallacy_and_mature_sciences
Ylikoski, Petri. Causal and Constitutive Explanation Compared. Erkenntnis 1975-. Vol. 78, Supplement 2: Causality and Explanation in the Sciences (2013), pp. 277-297. Available at:
http://www.researchgate.net/publication/257560716_Causal_and_Constitutive_Explanation_Compared
Comments
Post a Comment