5.6: Umwelten, Affordances, and Enactive Perception

The situatedness of an agent is not merely perception; the nature of an agent’s perceptual apparatus is a critical component of situatedness. Clearly agents can only experience the world in particular ways because of limits, or specializations, in their sensory apparatus (Uexküll, 2001). Ethologist Jakob von Uexküll coined the term umwelt to denote the “island of the senses” produced by the unique way in which an organism is perceptually engaged with its world. Uexküll realized that because different organisms experience the world in different ways, they can live in the same world but at the same time exist in different umwelten. Similarly, the ecological theory of perception (Gibson, 1966, 1979) recognized that one could not separate the characteristics of an organism from the characteristics of its environment. “It is often neglected that the words animal and environment make an inseparable pair” (Gibson, 1979, p. 8).

The inseparability of animal and environment can at times even be rooted in the structure of an agent’s body. For instance, bats provide a prototypical example of an active-sensing system (MacIver, 2008) because they emit a high-frequency sound and detect the location of targets by processing the echo. The horizontal position of a target (e.g., a prey insect) is uniquely determined by the difference in time between the echo’s arrival to the left and right ears. However, this information is not sufficient to specify the vertical position of the target. The physical nature of bat ears solves this problem. The visible external structure (the pinna and the tragus) of the bat’s ear has an extremely intricate shape. As a result, returning echoes strike the ear at different angles of entry. This provides additional auditory cues that vary systematically with the vertical position of the target (Wotton, Haresign, & Simmons, 1995; Wotton & Simmons, 2000). In other words, the bat’s body—in particular, the shape of its ears—is critical to its umwelt.

Passive and active characteristics of an agent’s body are central to theories of perception that are most consistent with embodied cognitive science (Gibson, 1966, 1979; Noë, 2004). This is because embodied cognitive science has arisen as part of a reaction against the Cartesian view of mind that inspired classical cognitive science. In particular, classical cognitive science inherited Descartes’ notion (Descartes, 1960, 1996) of the disembodied mind that had descended from Descartes’ claim of Cogito ergo sum. Embodied cognitive scientists have been strongly influenced by philosophical positions which arose as reactions against Descartes, such as Martin Heidegger’s Being and Time (Heidegger, 1962), originally published in 1927. Heidegger criticized Descartes for adopting many of the terms of older philosophies but failing to recognize a critical element, their interactive relationship to the world: “The ancient way of interpreting the Being of entities is oriented towards the ‘world’ or ‘Nature’ in the widest sense” (Heidegger, 1962, p. 47). Heidegger argued instead for Being-in-the-world as a primary mode of existence. Being-in-the-world is not just being spatially located in an environment, but is a mode of existence in which an agent is actively engaged with entities in the world.

Dawson, Dupuis, and Wilson (2010) used a passive dynamic walker to illustrate this inseparability of agent and environment. A passive dynamic walker is an agent that walks without requiring active control: its walking gait is completely due to gravity and inertia (McGeer, 1990). Their simplicity and low energy requirements have made them very important models for the development of walking robots (Alexander, 2005; Collins et al., 2005; Kurz et al., 2008; Ohta, Yamakita, & Furuta, 2001; Safa, Saadat, & Naraghi, 2007; Wisse, Schwab, & van der Helm, 2004). Dawson, Dupuis, and Wilson constructed a version of McGeer’s (1990) original walker from LEGO. The walker itself was essentially a straight-legged hinge that would walk down an inclined ramp. However, the ramp had to be of a particular slope and had to have properly spaced platforms with gaps in between to permit the agent’s legs to swing. Thus the LEGO hinge that Dawson, Dupuis, and Wilson (2010) built had the disposition to walk, but it required a specialized environment to have this disposition realized. The LEGO passive dynamic walker is only a walker when it interacts with the special properties of its ramp. Passive dynamic walking is not a characteristic of a device, but is instead a characteristic of a device being in a particular world.

Being-in-the-world is related to the concept of affordances developed by psychologist James J. Gibson (Gibson, 1979). In general terms, the affordances of an object are the possibilities for action that a particular object permits a particular agent. “The affordances of the environment are what it offers the animal, what it provides or furnishes, either for good or ill” (p. 127). Again, affordances emerge from an integral relationship between an object’s properties and an agent’s abilities to act.

Note that the four properties listed—horizontal, flat, extended, and rigid—would be physical properties of a surface if they were measured with the scales and standard units used in physics. As an affordance of support for a species of animal, however, they have to be measured relative to the animal. They are unique for that animal. They are not just abstract physical properties. (p. 127)

Given that affordances are defined in terms of an organism’s potential actions, it is not surprising that action is central to Gibson’s (1966, 1979) ecological approach to perception. Gibson (1966, p. 49) noted that “when the ‘senses’ are considered as active systems they are classified by modes of activity not by modes of conscious quality.” Gibson’s emphasis on action and the world caused his theory to be criticized by classical cognitive science (Fodor & Pylyshyn, 1981). Perhaps it is not surprising that the embodied reaction to classical cognitive science has been accompanied by a modern theory of perception that has descended from Gibson’s work: the enactive approach to perception (Noë, 2004).

Enactive perception reacts against the traditional view that perception is constructing internal representations of the external world. Enactive perception argues instead that the role of perception is to access information in the world when it is needed. That is, perception is not a representational process, but is instead a sensorimotor skill (Noë, 2004). “Perceiving is a way of acting. Perception is not something that happens to us, or in us. It is something we do” (p. 1).

Action plays multiple central roles in the theory of enactive perception (Noë, 2004). First, the purpose of perception is not viewed as building internal representations of the world, but instead as controlling action on the world. Second, and related to the importance of controlling action, our perceptual understanding of objects is sensorimotor, much like Gibson’s (1979) notion of affordance. That is, we obtain an understanding of the external world that is related to its changes in appearance that would result by changing our position—by acting on an object, or by moving to a new position. Third, perception is to be an intrinsically exploratory process. As a result, we do not construct complete visual representations of the world. Instead, perceptual objects are virtual—we have access to properties in the world when needed, and only through action.

Our sense of the perceptual presence of the cat as a whole now does not require us to be committed to the idea that we represent the whole cat in consciousness at once. What it requires, rather, is that we take ourselves to have access, now, to the whole cat. The cat, the tomato, the bottle, the detailed scene, all are present perceptually in the sense that they are perceptually accessible to us. (Noë, 2004, p. 63)

Empirical support for the virtual presence of objects is provided by the phenomenon of change blindness. Change blindness occurs when a visual change occurs in plain sight of a viewer, but the viewer does not notice the change. For instance, in one experiment (O’Regan et al., 2000), subjects inspect an image of a Paris street scene. During this inspection, the color of a car in the foreground of the image changes, but a subject does not notice this change! Change blindness supports the view that representations of the world are not constructed. “The upshot of this is that all detail is present in experience not as represented, but rather as accessible” (Noë, 2004, p. 193). Accessibility depends on action, and action also depends on embodiment. “To perceive like us, it follows, you must have a body like ours” (p. 25).