Tutorial on Embodiment
2.1. The Study of Intelligence
In this section, we will neither attempt to define or characterize intelligence, nor specify the right way to study it. Instead, we will only point out two important aspects. First, we want to emphasize the potential of the synthetic approach to studying intelligence: "understanding by building". Second, we want to draw attention to the frame of reference problem: too often, we automatically attribute intelligence to a certain behavior, without realizing that the mechanism that brings it about may be surprisingly simple.
The synthetic approach
"We distinguish between analytic and synthetic approaches. The analytic approach is universally applied in all empirical sciences. Typically, experiments are performed on an existing system, a human, a desert ant, or a brain region, and the results are analyzed in various ways. Often the goal is to develop a model to predict the outcome of future experiments. By contrast, the synthetic approach works by creating an artificial system that reproduces certain aspects of a natural system (Fig 2.1.1.). This is another important function of models. Rather than focusing on producing the correct experimental results, that is, the correct output, we can try to reproduce the internal mechanisms that have led to the particular results. In a memory experiment, we could predict, say, the number of items recalled, based on a statistical model. Alternatively we could try to model the memory processes themselves. An ethologist may want to predict where an ant path will be formed. Again, he can use statistical modeling, but he can also attempt to model the behavioral rules by which the ants interact with the environment and with each other. Such models are typically computer models that, when run, are expected to reproduce the experimental results. The focus of interest shifts from reproducing the results of an experiment, although that is still an important aspect, to understanding why the results come about. This kind of approach is called synthetic modeling and is extremely productive. It is at the core of the discipline embodied cognitive science. Such an approach can be characterized as "understanding by building"." (Pfeifer and Scheier 1999, p. 21-22)
Fig 2.1.1. (a) (b) (c)
"The synthetic methodology: scientist trying to understand ant behavior. (a) Directly mapping observed behavior onto an internal representation. (b) Using this representation to control walking in a robot fails. (c) Applying a much simpler model to control walking in a robot. This latter, simpler model seems more plausible as an explanation of the originally observed behavior."(cartoon by S. Iwasawa from Pfeifer and Bongard, 2007)
Frame of Reference problem
"Agents can exploit physical laws even if they are not aware of them. Intelligence, in this sense, is not so much a property of an agent or of the brain or of evolution, but rather resides in the eye of the beholder, so to speak, who observes the exploitation. This leads us to the next general problem that the study of intelligence raises: the so-called frame-of-reference problem. The frame-of-reference issue, which is concerned with the perspectives that we can adopt when observing or designing agents, implies that we must be very clear about what we are observing and how we interpret what we observe. The initial inspiration for this line of thought comes from Herb Simon's seminal book The Sciences of the Artificial, in which he introduced the anecdote of an ant walking along a beach (Simon, 1976). He argued that from an observer's point of view, the ant describes a complex path because it walks around puddles, rocks, twigs, and pebbles. However, from the point of view of the ant, the mechanisms that bring about this behavior might in fact be quite simple, such as "if obstacle on right then turn left" or "if obstacle on left then turn right," and "go straight." The final path of the ant emerges from its interaction with the environment; in this case, a beach. The ant knows nothing about puddles, pebbles, and twigs but still manages to find its way around quite well (see also Pfeifer and Scheier, 1999)." (Pfeifer and Bongard 2007, p. 72)
A good illustration of the frame of refrence problem is the following video 2.1.1. It basically shows triangles and circles moving around. However, people tend to attribute their own emotional state to these "creatures" and, consequently, make a story with human-like characteristics (The Oxford ethologist David McFarland called it "anthropomorphization: the incurable disease.").
Video 2.1.1: A movie from 1944 used to study the activation of anthropomorphic descriptions when watching moving geometric figures. (Heider and Simmel 1944)
Pfeifer, R. & Bongard, J. C. (2007), How the body shapes the way we think: a new view of intelligence, MIT Press, Cambridge, MA.
Pfeifer, R. & Scheier, C. (1999), Understanding Intelligence, MIT Press.
Simon, H. A. (1976). Administrative behavior: A study of decision-making processes in administrative organization. 3rd edition. New York: Free Press.