Now that we know that a cellular automaton can support a programmable computer, suppose we exercise the imagination just a little by assuming that the computer in the cellular automaton can send out glider-like probes that will bounce back from structures within the environment of the cellular automaton, returning to the computer like reflected radar signals, supplying the computer with information about its environment. The figure is a schematic of what this would look like on our desktop computer monitor.
Let us call the computer UTM-2 (the dash is actually a minus sign, for a reason that will become clear shortly) and suppose that it is programmed to use the information from the returning probes to build up a model of its environment in its memory, updating that model as the environment changes. In effect, UTM-2 is “aware” of its surroundings in the same way that, for instance, a driverless car is aware of its surroundings on the highway. We might say that, within the “space” of the Game of Life, UTM-2 is sentient, although not necessarily self-aware. Let us call the space containing UTM-2 “space -2”.
However, stretching our imagination only just a little further, let us include UTM-2 itself as part of the environment that it explores. This might not be such a tall order, since UTMs implemented in the Game of Life extend over vast numbers of cells, and we are only asking the probe to make low-resolution passes over the environment – just enough to pick out the different structures of the UTM.
One such structure that UTM-2 should be able to pick out is the internal model that it is making of its environment. So now we have a UTM that is not only aware of its environment: it is (albeit imperfectly) aware of the model that it is constructing of its environment – we might say that it is aware of itself! It may well be that such a feedback loop – i.e., input from the environment (which includes the observer) being compared with a similarly constructed model made at an earlier time – is necessary for the observer to be self-aware.
(Of course, self-awareness must entail more than just including oneself in the model one makes of one’s environment. For instance, for you, self-awareness, as you probably think of it, will include the model that you make of the model that other people have constructed of you!)
With a little licence, we can imagine that our desktop computer runs a Game of Life in which a UTM sends out probes that are reflected back by structures in the environment, enabling the UTM to use this reflected information continually to update a model of its environment. The environment can include itself, including the very model that it is updating, so that the UTM could be said to be “self-aware” in a crude sense. The label “space -2” is explained in the text.
We can even make the case that UTM-2 will conclude that it, and therefore its environment, exists! (Our case might start with that most famous utterance of René Descartes: I think, hence I am – or should that be I compute, hence I am ? – although some philosophers would regard that approach as stepping into philosophical quicksands, because, they argue, the use of the word I in Descartes’ dictum already presupposes the existence of “I” – it is a self-fulfilling prophecy!) Anyway, from the perspective of the space of the Game of Life in which UTM-2 operates, UTM-2, and everything else in its environment, is real. Reality within that space is the complete mathematical structure of that space.
The Game of Life can be thought of as a vertical stack of board positions, each subsequent position being generated from the rules of the game. Illustrated here is a snapshot from the Game of Life showing a complete cycle of a glider’s progress, taken from the image in Step 2.4.
A property of the Game of Life is that, once you decide on the configuration of the board at the beginning of the game – that is, the positions of all the live cells in the two-dimensional array of cells – then the configurations of the board at all subsequent time-steps are completely determined from the rules of the Game of Life, and which can be encapsulated as a mathematical structure. As an illustration, take the glider shown in Step 2.4, reproduced here as a vertical stack of time-steps. As before, the time-steps are labelled h, but this time we set the bottom time-step to zero, and so on upwards, so that the configuration at the beginning of the game, where h = 0, is the position of the live cells shown on the bottom slice.
The effect of this is that the only input required to run our Game of Life is its algorithm and the starting configuration of the board. Once Rendell had determined the positions of the live cells at h = 0 in his UTM, everything else followed according to our algorithm. So, in our slightly fanciful implementation of UTM-2 with its bouncing probes, once the starting position is fixed, so is the whole game: it is encapsulated by a monolithic structure like the one for the glider, only with more cells – effectively a little block universe. We can define the essence of the Game of Life as a mathematical structure:
where State (p, q, h + 1) is the state of the cell (the state is either 1 if the cell is “live”, or 0 if the cell is “dead”) at matrix position p, q and time step h + 1. N (p, q, h) is the number of live neighbours around the cell located at co-ordinates p, q at time step h. State (p, q, 0) is the configuration specified for the starting position of the Game (at h = 0).
I’m not expecting you to check this! The main reason for writing it down is to show that the essence of the pattern doesn’t have to involve grid-like diagrams. The double summation (the two Greek capital sigmas: Σ Σ) assigns values to triplets of numbers (p,q,h), and these in turn are used in assigning values to other triplets formed by incrementing the value of h. If you were a being in a one-dimensional universe, you would see the output of the equation as one long stream of numbers, because you wouldn’t have any conception of a higher dimension. We, however, can easily visualize the underlying pattern of number triplets as positions on a three-dimensional grid as in the figure, but that is a matter of convenience, not necessity. Indeed, the self-aware UTMs are interacting within a two-dimensional world, and the third dimension, represented by h, is more of an analogue of time to them. The mathematical structure is not just telling you where things are; it is telling you how these things are changing. If the output were in quintuplets of numbers instead of triplets, then we ourselves would have trouble visualizing the underlying pattern.
Notice, by the way, that the two equations are not themselves the mathematical structure: the mathematical symbols used, for example, were merely invented by people, whereas the mathematical structure is completely detached from any such cultural baggage. The equations are just our attempt to represent the underlying pattern in a slightly more detached way.
So, to be clear, we are saying that UTM-2, as part of this mathematical structure, is sentient and self-aware, on its own terms. The complete Game of Life is a mathematical structure that can include substructures that are aware of the reality of the space in which they exist!
(Note that, when a mathematical structure such as the moving glider is part of a larger mathematical structure, the Game of Life, then we also refer to it as a mathematical substructure.)