For our investigation we have restricted the world to a stage

= Goals of the research, and problems that must be addressed

The overriding go= al for the research is to investigate language, language use, and language comprehension in such a way that it is compatible with evolution and hopefu= lly also provides insight into language-based learning. Beyond that, it would be nice to m= ake a contribution to the mind body discussion.

We have chosen to= tackle the problem from an information processing perspective, with the use of computer simulation to illustrate and clarify issues. Since we are operating inside a bl= ack box, where little direct empirical verification is possible, we have to find new methods for analysis and for verification.

= The need for compression

For muscle contro= l we estimated a minimum of 7500 bits for each frame, but with multiple muscles = and redundancy it would probably be more than 100 times this estimate, i.e. alm= ost a megabit. For visual input we estimated at least 8 megabits for each frame – corresponding to a sin= gle glance or view. The two-direc= tional interface between the brain and the world seems to be very busy. A 10 minute segment (at a low 15 f= rames per second) would be 9000 frames or 81,000,000,000 bits or over 10 gigabytes. Compression seems = called for.

= The need for data compatibility

The information f= lowing through the two subsystems as presented above is not compatible. The first subsystem receives 2-dim= ensional geometric information about the position of the bones that are not obscured= by other bones. This information= is relative to the stage and to the point of view from the eye of the observer. The second subsyste= m also has geometric information, but it is expressed as the relative change of an= gle for each bone and possibly in more than one dimension if a joint has multip= le degrees of freedom such as the hip and shoulder joints.

There is also a p= roblem of timing. Let us assume that= the actor has turned his head so that he can see some of his own bones. Let us further assume that at time= 1 the brain-output has sent change-angle information. This change should be implemented by the relevant muscles by time2.<= span style=3D'mso-spacerun:yes'> Assuming a small perceptual delay,= the new position should be registered by the brain at time3. If the perception information feed= back is used as position feedback, the feedback would be delayed, which commonly leads to an oscillating response or a very slow response.=

= The need for action coordination

Up to this pint w= e have not dealt with the coordination of action but posited that movement informa= tion for each bone is sent separately over distinct neural paths. Even in a simple, relatively static pose, the bones have to cooperate. There are many bone positions that would cause the actor to fall ove= r or look totally absurd (c.f. Monty Python, the Ministry of Silly Walks). Similarly, bone positions have to = be coordinated over time, for instance in walking. Beyond that, much action is purpos= ive or goal-oriented. Again, using introspection, most of us do not know motion about joints but think of moti= on at a more abstract level including several bones and a succession of moveme= nts in a timed sequence. There are times, such as when learning to play golf, that we try to analyze our own motions in much more detail – but we generally find it difficult.