2002 WhenisaCognitiveSystemEmbodied

(Riegler, 2002) ⇒ Alexander Riegler. (2002). “When is a Cognitive System Embodied?.” In: Cognitive Systems Research Journal, 3(3). doi:10.1016/S1389-0417(02)00046-3

Subject Headings:

Notes

Cited By

Quotes

Author Keywords

Complexity; Constructivism; Design; Embeddedness; Representation; Teleonomy; Understanding

Abstract

For cognitive systems, embodiment appears to be of crucial importance. Unfortunately, nobody seems to be able to define embodiment in a way that would prevent it from also covering its trivial interpretations such as mere situatedness in complex environments. The paper focuses on the definition of embodiment, especially whether physical embodiment is necessary and/or sufficient for cognitive systems. Cognition is characterized as a continuous complex process rather than ahistorical logical capability. Furthermore, the paper investigates the relationship between cognitive embodiment and the issues of understanding, representation and task specification.

Introduction

We all are living beings. As such, if we look around we perceive a world full of different shapes, colors, sounds, smells, etc. All these modalities are not in merely chaotic disorder. By using our cognitive abilities, we find sense in the world. We can identify objects, and distinguish them from one another. We also can manipulate things, thus changing relationships between them. Making sense of the world helps us to survive. This all seems very self-evident and trivial to us. But is it? We have not had most of our capabilities at birth. For a newborn baby, almost nothing is obvious. There is an ongoing (and actually never-ending) process of cognitive development which makes us what we are.

But what about our pets, what about animals? Do they just experience a chaotic disorder of color and noise? Apparently they donft either. Observing animals, either as an amateur or an ethologist, reveals their remarkable capabilities. So is there a difference? Obviously, most animals do not talk to each other, and perhaps all have problems in solving mathematical equations. We conclude, there must be a gradual difference between humans and dogs, between cats and amoebas. One thing they all have in common: strategies to survive, or more precisely, to live. Strategies to be developed and executed need a cognitive apparatus of some sort. In other words, creatures survive in their environments by using their cognitive abilities which is in turn shaped by the interaction Alexander Riegler: When Is a Cognitive System Embodied? Page 2 with the environment. As Rodney Brooks puts it, gIntelligence is determined by the dynamics of interaction with the world.h (Brooks 1991, p. 585) . This is, in a nutshell, the motivation for developing the concept of embodiment, which entered cognitive science and related disciplines in the 1980s.

In this paper we will focus on the definition of embodiment, especially whether physical embodiment is a necessity for cognitive systems, and its relationship to the issues of understanding, representation and task specification. We will make use of the idea that cognition is a continuous complex process rather than an ahistorical logical capability. The goal is to come up with a clearer idea of the role of embodiment for cognitive systems.

Cognition

The complexity of cognition.defined in terms of behavioral repertoire that enables adequate compensation of perturbations from the environment.is different for different individuals: As we have seen in the introduction, there is a wide range between simplest forms of cognition in simple lifeforms and humanlike cognition.1 If we trust in the idea of evolution we have to ask for the mechanisms that evolved the wide range of different cognitive apparatus over time.

When we speak about evolution, we implicitly assume a gradual adaptation of species towards their environment. In 1941, Konrad Lorenz, for example, stated that the horsefs hoof is a representation (Abbild) of the steppe, the body form of the dolphin is the incarnation of knowledge about laws of aerodynamics in water, etc. (Lorenz 1982) 2. Likewise, cognitive capabilities seem to have emerged in resonance to the structure of the environment (resulting in eanschauungsformenf in the terminology of Lorenz). We are intelligent because our environment has been challenging enough to select for the smart guys. The idea that the physical world influences the behavior of an agent (rather than being fed with instructions from the programmer) is commonly referred to as gsituatednessh (Pfeifer & Scheier 1999). Embodied beings deal with the world, and their cognitive capabilities emerge out of this interaction.

How can we conceive of the interaction between system and environment as the engine of cognitive complexification? Taking Lorenz’s statement further, cognitive capabilities are also a reflection of the environment 4. This idea leads directly to the proposal that Karl Popper labeled the ‘bucket theory of mind’ (1979). According to this view, “there is nothing in our mind which has not entered through our senses”. Thus, animals and humans are cognitive buckets which get filled over time. While this concept has much common sense attractiveness, it quickly runs into problems such as the frame problem (Dennett 1984). How do you feed your cognitive apparatus with the facts of the ‘outside world’? How do you formally specify what changes in your environment and what remains constant? As each fact in the world is potentially connected with any other piece of fact we would need to update the content of our bucket each time there is a change in our environment. The wrong assumption here is that the world is a collection of facts that could be arbitrarily combined with each other. Even if we managed the combinatorial complexity, a question would remain: What is a fact? Entities in our perception don’t come labeled (Franklin 1995). If we look at a tree, we know that it is a tree – but how has this meaning emerged (cf. the symbol grounding problem, Harnad 1990)?

Scientists aware of these tricky and annoying problems, which made artificial intelligence an unreachable goal in the past, proposed the concept of embodiment as a solution. It refers to the idea that gintelligence cannot merely exist in the form of an abstract algorithm but requires a physical instantiation, a bodyh (Pfeifer & Scheier 1999). Through embodiment, symbols get physically grounded, and such meaning is defined through interaction with the world. However, the importance of physical embodiment has been questioned. Oren Etzioni (1993) criticizes that building robots which interact with the complexity of the ereal worldf are not the exclusive road to embodied cognition. Operating systems such as UNIX too provide a sufficiently complex and hence challenging environment for cognitive agents. Tom Quick et al. (1999) propose (non-optimal) information retrieval agents on the internet to be candidates of embodiment. Their ephenomorphf program is based on the idea that its relationship to the internet is analogous to the one of the bacterium escheria coli to its environment. Stan Franklin (1997) discusses structurally coupled with their environmenth (Franklin 1997, p. 500, my emphasis). These and other authors share the common view that embodiment does not necessarily mean physically embodied.

…

References

,

	Author	volume	Date Value	title	type	journal	titleUrl	doi	note	year
2002 WhenisaCognitiveSystemEmbodied	Alexander Riegler			When is a Cognitive System Embodied?				10.1016/S1389-0417(02)00046-3