Why Computers Can’t Feel Pain

Cf. Minds and Machines, 4: 4, ‘What is Computation?’, November 1994.

A ‘maximal’ state is a total state of the system, specifying the system’s physical makeup in absolute detail.

Chalmers (1996) observes, “Even if it [the claim that ‘every open physical system is a realisation of every abstract finite state automaton’] does not hold across the board (arguably, signals from a number of sources might cancel each other’s effects, leading to a cycle in behaviour), the more limited result that every non-cyclic system implements every finite-state automaton would still be a strong one”.

Clearly there may be other state transition sequences that have not emerged in this execution trace. To circumvent this problem and fully implement an inputless FSA by an infinite state [counter] system, Chalmers posits the system with an extra dial—a sub-system with an arbitrary number of states, [c _{[dial-state, counter-state]}]. Now, associate dial-state [1] with the first run of the FSA. The initial state of the counter machine will thus be [c _{[1, 1]}] and we associate this with an initial state of the FSA. Next associate counter states [c _{[1, 2]}], [c _{[1, 3]}] with associated FSA states using the Putnam mapping described earlier. If at the end of this process some FSA states have not come up, we choose a new FSA state, [C], increment the dial of the counting machine to position [2] and associate this new state [c _{[2, 1]}] with [C] and proceed as before. By repeating this process all of the states of the FSA will eventually be exhausted. Then, for each state of the inputless FSA there will be a non-empty set of associated counting machine states. To obtain the FSA implementation mapping we use Putnam’s mapping once more and the disjunction of these states is mapped to the FSA state as before. Chalmers remarks, “It is easy to see that this system satisfies all the strong conditionals in the strengthened definition of implementation [above]. For every state of the FSA, if the system is (or were to be) in a state that maps onto that formal state, the system will (or would) transit into a state that maps onto the appropriate succeeding formal state. So the result is demonstrated.” (Chalmers 1996, p. 317). However this extension is not required for the argument developed herein.

“To see the triviality, note that the state-space of an inputless FSA will consist of a single unbranching sequence of states ending in a cycle, or at best in a finite number of such sequences. The latter possibility arises if there is no state from which every state is reachable. It is possible that the various sequences will join at some point, but this is as far as the ‘structure’ of the state-space goes. This is a completely uninteresting kind of structure, as indeed is witnessed by the fact that it is satisfied by a simple combination of a dial and a clock. (ibid., p. 318).

NB. It is central to the computationalist underpinning of cronos that its putative conscious states are not contingent upon it physically interacting with a physical environment; in personal communication, Prof. Holland envisaged a possible follow up project in which the entire cognitive architecture of cronos and its environment are entirely implemented in software, in a large scale virtual reality simulation.

“Suppose that a system exists whose activity through a period of time supports a mode of consciousness, e.g. a tickle or a visual sensum. The supervenience thesis tells us that, if we introduce into the vicinity of the system an entirely inert object that has absolutely no causal or physical interaction with the system, then the same activity will support the same mode of consciousness. Or again, if the activity of a system supports no consciousness, the introduction of such an inert and causally unconnected object will not bring any phenomenal state about … if an active physical system supports a phenomenal state, how could the presence or absence of a causally disconnected object effect that state?” (Maudlin 1989).

A ‘conditional branch’ instruction is an instruction in a computer program of the form, “IF (TEST IS TRUE) THEN GOTO {statement sequence A} ELSE GOTO {statement sequence B}”.

The $ sign indicates a hexadecimal number; i.e. a number to the base 16; dgit range is [0 .. 9 A .. F], hence hexadecimal $FF is 15 × 16 + 15 = 255 (decimal).

Clearly, if the phenomenal experience of robot D differed from robot A, then the putative phenomenal states of a robot will always be contingent upon the particular type of compiler used by the roboteer (not on the semantics of actual program he or she wants to compile).

Objection raised by a member of the audience at the presentation of this paper at the 2006 ‘Computers and Philosophy’ conference, Laval, France.

Although it is true that as the complexity of the logical system increases, the number of consistent computational functions that can be assigned to it diminishes, it remains the case that its computational properties will always be relative to the threshold logic value used. The ‘physical-state’ ⇒ ‘computational-state’ mapping will always co-determine the ‘logical-function’ that the physical computational system instantiates.