A 10 Minutes Introduction to Embodied Cognition
// December 12th, 2012 // Artificial Intelligence
What is cognition?
Basically, it is a group of mental processes.
Cognition requires:
- Perception
- Attention
- Anticipation
- Reasoning
- Learning
- Inner Speech
- Imagination
- Memory
- Emotions
- Planning
- Pain and Pleasure
Most cognitive scientists view cognition as something that is computational.
Cognition = Computational System
By computational system they mean a system that manipulates symbols, in the sense that symbols without meaning are manipulated by means of the application of rules, generating by hence new symbols and conclusions.
How a living system gets the symbols?
A sensor gathers meaningful data. This data is converted into symbols. Then the brain uses the symbols to generate a (symbolic) response. The response is translated into meaningful action data that is executed by the actuators of the living system.
This explanation is perfect for artificially intelligent systems developers because it indicates that the brain doesn’t need a body. Then the scientists can concentrate on generating intelligence in any physical system that allows manipulation of symbols, and forget about the hardware.
There is though a small problem. How do the symbols acquire and release their meaning? The process by which meaningful data is translated into symbols, and symbols are translated back into meaningful data is called reification and up to date, nobody knows how it is performed (it is called the grounding problem) … at least in a non-embodied cognitive framework
Let’s make then an hypothesis:
meaning arises from the nature of the body
What this means?
- Living things generate a basic set of meaningful concepts based on their interaction with the world
- More complex concepts can be generated by applying metaphores over previous concepts.
This approach requires of a body to generate intelligence. That is the approach of the embodied cognition paradigm, and has the following implications:
- Conceptualization: the properties of an organism body constrain the concepts it can acquire (this has big implications for making artificial systems understand)
- Replacement: interaction with the environment replaces the need for representations (this has big implications in the number of resources needed to create an A.I.)
- Constitution: body is constitutive of cognitive processes rather than causal (this has big implications in how perception should be made in artificial systems)