(1) The notion of an "extended mind" -- with mental states (i.e., felt states) 'distributed' beyond the narrow bounds of the individual brain – is not only as improbable as the notion that the US government can have a distributed migraine headache, but arbitrary.

(2) "Cognition" -- if it is simply defined as the ability to do the kinds of things that cognizers like us can do, plus the underlying functional mechanisms for doing them -- can be arbitrarily defined to be as wide or as narrow as we like.

(3) Vagueness about the nature, locus and scope of cognizing leads to a dissociation of 'cognitive states' from mental states. However, their co-occurrence had been our only basis for distinguishing cognitive performance capacity from other capacities and functionality (animate or inanimate, narrow or wide).

(4) If cognitive states are indeed not mental states, it follows that "cognitive technology" is not just something used by cognizers, but a functional part of the cognitive states themselves, because the boundary between user and tool disappears, and cognitive states become merely instances of functional states in general.

(5) We then do not need the terms "cognitive" and "distributed cognition" at all, and can just talk about relatively complex and wide or narrow functional states, leaving it a coincidence and mystery (at least at this stage) that every single case of what we used to call 'cognitive' also happened to be mental.

(6) A way to resolve this is to accept that only mental states are cognitive states, that cognition is only narrow, and that the only place it is "distributed" is within a single cognizer's brain.

(7) The only kind of 'technology' that might really turn out to be intrinsically cognitive, rather than just being a tool used by cognizers, would be a robot that could pass the Turing Test (TT) -- because such a TT-scale robot would almost certainly have mental states, and hence it would be a cognizer in its own right.

(8) Whatever distributed activity was going on within the functional mechanism generating such a TT robot's performance capacity would then indeed be a case of distributed cognition (exactly as the distributed activity within our own brains is distributed cognition) – even if not all the components of its generating mechanism were located inside the robot's head.

(9) The 'cognitive technology' used by such a TT robot, however, would still not be part of its distributed cognitive (hence mental) state, just as it is not a part of ours.

(10) Nor would a group of such TT robots, interacting and collaborating, be a case of distributed cognition; it would merely be a case of collaborative cognition among individual (narrow) TT-robot cognizers, just as it is in the case of a group of collaborating human cognizers.

(11) Cognitive technology does, however, extend the scope and power of cognition, exactly as sensory and motor technology extends the scope and power of the bodily senses and movement.

(12) Just as we can see further with telescopes, move faster with cars, and do more with laser microsurgery than we can do with just our unaided hands and heads, so we can think faster and further, and do more, with language, books, calculators, computers, the web, algorithms, software agents, plus whatever is in the heads of other cognizers.

(13) Both sensorimotor technology and cognitive technology extend our bodies' and brains' performance capacities as well as giving us the feeling of being able to do more than just our bodies and brains alone can do.

(14) Sensorimotor and cognitive technology can thus generate a perceptual change, rather like virtual reality (VR), making us feel a difference in our body image and causal power (perhaps not unlike what the physical metamorphosis from caterpillar to butterfly might feel like, as one sensed one's newfound somatic capacity to fly).

(15) This change in perceived body image is indeed a change in mental state; but although its distal inputs and outputs certainly extend wider than the body (as all sensory inputs and all motor outputs do), the functional mechanism of that altered mental state is still just proximal -- skin and in – exactly as when it is induced by VR technology.

(16) Hence, although sensorimotor and cognitive technology can undeniably extend our bodies' sensorimotor and cognitive performance powers in the outside world, only their sensorimotor input and output contact points with our bodies are part of our cognitive (= mental) state, not the parts that extend beyond.

(17) Perhaps it could be otherwise too, as in the case of a hypothetical TT-robot whose generating mechanism is indeed partly located outside its body: Maybe parts of our brain could be removed and still functionally integrated with the rest wirelessly, through telemetry or some other action at a distance: But that would just be a widened, spatially distributed body.

(18) The resultant distributed cognitive state would still not be the same thing as considering a telescope, car, library or calculator as parts of a distributed cognitive state (for either a human or a TT robot): Those would still just be parts of the sensorimotor I/O to and from the cognizer's body.

(19) We are not aware of the generating mechanism underlying our cognitive capacity, however, only of its outcome: Hence retrieving a word from memory or retrieving a word via a Google search feels much the same to us.

(20) Does the fact that cognizing is a conscious mental state, yet we are unconscious of its underlying functional mechanism, mean that the underlying functional mechanism could include Google, Wikipedia, software agents and other human cognizers' heads after all? That question is left open for the reader.

(21) The worldwide web, a distributed network of cognizers, digital databases and sofware agents, has become our 'Cognitive Commons,' in which cognizers and cognitive technology can share cognizing anytime and anywhere, and interact globally with a speed, scope and degree of interactivity that yield distributed cognizing with performance powers inconceivable within the scope of individual cognition.

(22) Such changes go beyond mere quantitative increase in efficiency and performance power. As we increase our use and reliance on cognitive technologies, they effect and modify how we cognize, how we do things and what we do. Just as motor technology extended our physical ability and modified our physical life, cognitive technology extends our cognitive ability and modifies our mental life.

Excerpted from:

Dror, I. and Harnad, S. (2009) Offloading Cognition onto Cognitive Technology. In: Cognition Distributed: How Cognitive Technology Extends Our Minds, John Benjamins.