Last visitors [hide]
1)   admin at Sep 22, 2014 [12:43]
2)   huangyn2j at Jan 17, 2013 [15:50]
3)   KevinORegan at Dec 28, 2010 [16:43]
4)   kevintesting at Jan 01, 1970 [00:00]
5)   test3jan2011 at Jan 01, 1970 [00:00]
6)   del at Jan 01, 1970 [00:00]
7)   randybosh at Jan 01, 1970 [00:00]
8)   StephanieSpino at Jan 01, 1970 [00:00]
9)   Humberto321 at Jan 01, 1970 [00:00]
10)   Marcossp at Jan 01, 1970 [00:00]

Key Concepts 2: Sensorimotor skill


Sensory input and motor output

Sensory input or stimulation is the signal provided by a system's sensors. Motor output is the signal sent to the effectors. The notions of "sensor" and "effector" are not trivial, and may depend on the point of view from which one analyses the interaction of the system with its environment. When a system is built for a purpose, then within that context, it is easier to define what are sensors and effectors. The notion of "signal" is also not trivial: do we mean the actual neural or electrical impulses, coded in the particular way they are coded? Or do we mean some intrinsic aspect of the input and output, independently of how it is coded. More thought is needed here...

Note in any case that we wish to use the term sensory input to refer to the (objective) information available to the system, given the properties of the sensors. In our terminology, we distinguish sensory input from what we call sensation, which has a phenomenal aspect, and from what we call perception which has both a phenomenal and a categorical aspect.
More on Sensory input and motor output?

A sensorimotor interaction

An interaction between a system and environment which involves the system activating its motor effectors and concomitantly receiving input from its sensory receptors. (Example: a baby waves its arms haphazardly and at the same time receives visual, auditory, tactile and olfactory input from its sensors. Another example: a broken thermostat haphazardly turns the heating on and off while measuring temperature variations in the room. Here the system is broken, but it is still having a sensorimotor interaction.
More on A sensorimotor interaction?

A sensorimotor skill

A sensorimotor interaction in which a system built for a purpose controls the interaction in a skillful way, that is, it can modify the input-output in a way that is in some sense adapted to one of its purposes. (Example, the baby moves its arms to grasp the approaching bottle; Another example: a thermostat operating properly turns the heating on and off as a function of the temperature in the room). (NB: the notion of "adapted" contains profound premisses about the viewpoint from which one is analysing the system. It seems to require that it make sense to say that the system is built for a purpose. The issue is tricky and needs careful consideration; there is the beginning of a discussion of this in O'Regan & Noƫ, BBS, 2002).
More on A sensorimotor skill?

Exercising a sensorimotor skill

Being in the process of executing a sensorimotor skill.
More on Exercising a sensorimotor skill?

Exercising mastery of a sensorimotor skill

In previous work I have used this phrase to emphasize that one is exercising the skill in the way it should be exercised, given the system's purpose, and not simply exercising a sensorimotor interaction haphazardly. But since the notion of skill seems to require the adapted, purposeful nature of the interaction, it is probably redundant to use the term "mastery". I will here assume that exercising a skill is by defnition "masterful".
More on Exercising mastery of a sensorimotor skill?

A sensorimotor contingency or law

The law that describes the regularities linking changes in effectors to resulting changes in sensory input when an agent exercises a sensorimotor skill. A problem: what do we define as "sensory input": should we take the information directly at the sensor output or after some degree of processing? Successive stages of brain processing following sensing will result in progressive loss in information. What do we need to use to define the law? The same question can be asked for effector output. It seems that one should consider the whole sensorimotor loop, going from the outside environment to sensors to brain to effectors, back to outside environment (or the other way), and consider as a "law" the properties of that loop that are available for categorization and manipulation by the cognitive processes of the system. More work needs to be done to clarify this issue.

Another problem: sensorimotor contingencies can be described using different codes — for example the law linking changes in auditory input as a function of distance from a sound source will be expressed in a different way depending on the units one uses to code the sound energy and the distance. We would like to retain in the notion of "law" only those aspects of the description of the sensorimotor skill which are independent of the code used to describe the law. The reason we require this is that we require that the only entities that the cognitive system of the agent has access to should be facts about the system's interaction with the environment. If we allowed the opposite, that is, if the reasoning done by cognitive system was reasoning about particular codes used somewhere in the system, then the system would have no natural metric to compare these entities, nor no way of making the link between these entities and experiences of other agents. (This argument seems to be related to Dennett's idea of heterophenomenology: a system has no privileged access to its own internal states).

Of course the code used to describe sensorimotor contingencies determines to some extent the possible variations in sensorimotor behavior that can be described, and may limit the calculations that the system's cognitive system can perform with regard to the law. More thought is needed to be clear on these problems.

Examples of sensorimotor contingencies: the sensorimotor contingency of seeing involves facts like, when you blink there is a big change in sensory input. When you move forward the sensory input changes in a special way (an expanding flow field). When you move your eyes, there are certain changes which are typical of seeing.
More on A sensorimotor contingency or law?

Being engaged in exercising a sensorimotor skill

The agent, who has a choice about it, is currently exercising a skillful sensorimotor interaction. The notion of being engaged requires having the possibility of not being engaged. The idea is that the system is invoking its cognitive ressources to exercise this particular sensorimotor skill, rather than that different one. Thus, it makes no sense to say that a thermostat itself is "engaged" in regulating the temperature in a room (it has no choice, no cognitive ressources to invoke). But consider an automatized home environment that can dim lights, regulate the heating, open and close blinds, etc. Then in winter, when the heating is turned on, it would make sense to say that the automatized home environment was engaged in regulating the furnace. But in summer the system would not be engaged in doing this, rather it would be engaged in operating the air conditioning.
More on Being engaged in exercising a sensorimotor skill?

Experiencing a raw sensory feel

Here we reach the culmination of our effort to make precise all the concepts underlying our definition of feel. We can now say that experiencing a raw sensory feel is: Having cognitive access to the fact that one is currently engaged in exercising a sensorimotor skill. We want this definition of experience to allow for the possibility that animals with even minimal cognitive capacities should have feels. But clearly the degree to which it makes sense to say an animal has feel depends on the degree to which it makes sense to say it can have cognitive access to something. Presumably a dog has some degree of cognitive access to the fact that it sees the cat. But presumably the fly has less cognitive access to the fact it is chasing another fly. The tic tac toe playing program has cognitive access to its move, but the move is not a sensorimotor skill, so one cannot say that the program "feels" it. What about if the tic tac toe machine had a robot arm which automatically made marks on the paper? Would it then "feel" the pencil it clenched in its robot fingers? Probably not, because it would have to have cognitive access to the skill of clenching. This would require there to be different modes of clenching and it would require the system to have a choice about clenching this way or that, and for it to be poised to make use of this clenching skill in its planning, decisions and communication. But presumably the system is not wired up to 'think' about the way it clenches, only about the actual moves it's making. It does not feel the pencil in its clench...
More on Experiencing a raw sensory feel?

Unconscious feel?

The definition of feel we are using expressly appeals to 'cognitive' and not 'conscious' access. This allows for a definition in which feels can be felt even by organisms who have no "selves", and even by organisms which have very limited cognitive capacities. In that sense therefore we could admit that in these cases there are unconscious feels. Furthermore, humans can also have unconscious feels under this definition: it suffices that a human be engaged in exercising a sensorimotor skill which is poised to affect his or her rational behavior, but without the human being aware of (i.e. having cognitive access to) this fact. Subliminal sensory stimulation as in the image flashed during the election campaign may be such an example.
More on Unconscious feel?

The phenomenal quality of raw sensory feel

The experienced or phenomenal quality of a raw sensory feel is constituted by the sensorimotor contingencies involved. This single idea is the crux of the sensorimotor approach to phenomenal consciousness. Several points should be noted about this definition.

First, we use the word "constituted" and not the word "caused". This is because we have defined feel in such a way that it is an entity for which it is nonsensical to speak of a cause. Thus, contrary to other approaches, in particular to approaches where some neural mechanism is assumed to cause feels, we say here that feels are not the kind of things that can be caused, because they are on the contrary constituted by a certain state of affairs concerning the observer's interaction with the world. The example of the softness of the sponge illustrates this clearly: the feeling of softness is not a kind of fluid that is generated anywhere or caused to be exuded by any mechanism. What we mean by feeling softness is: having cognitive access to the fact that we are doing the kind of thing we do when we press on the sponge.

A second point is that we are claiming that it is not the feel itself, but the quality of the feel that is constituted by the sensorimotor contingency. This is important, because it is not sufficient for a system to be engaged in exercising a sensorimotor skill for there to be a feel. There must additionally be cognitive access to this fact.

A third point is that because we are identifying the experienced quality of a feel to a sensorimotor contingency, we have the basis for explaining the similarities and differences between feels. This provides an advantage over other theories which invoke some causal mechanism: such theories must then explain why and how the particular mechanism produces the particular pattern of similarities and differences in feels that exist.

It is this last point which gives us such strong leverage to overcome the mystery of feel and answer The Important Questions. Forfeiting Question 1 on the neural origin of feel, we no longer nourish the hope that the future will turn up some arcane and newfangled mechanisms to explain the special characteristics of phenomenal experience. Instead we show how it is precisely the nature of sensorimotor skills which accounts for these unique properties. (see corporality and alerting capacity).
More on The phenomenal quality of raw sensory feel?


The "feel" that is experienced as a consequence of an agent's interacting with sensory stimulation. Unlike sensory input which is an objective physical entity, we take sensation to be the experience or feel associated with sensory stimulus.
More on Sensation?


The interpretation or semantic category that the agent gives to Sensations. NB having perceptions requires that the agent possess sufficient cognitive capacities to be able to "interpret" the sensations.
More on Perception?

Contributors to this page: KevinORegan , Melany , Brandon , JessMartone , dollievaruna , gregory , ALLISON and admin .
Page last modified on Thursday 04 of May, 2006 [11:53:54 UTC] by KevinORegan.

Free Hit Counter