5.5.5 Third review of the requirements for information processing

We introduced an "inner language" as a method for converting perception into action.

The grammatical forms and the vocabulary of this 'inner language' are very much open to speculation and may be very idiosyncratic. Some aspects of the semantics are required, such as the capability to describe visual object and to command (instruct) future action.

• As grammar, we conjecture a "subject verb object" structure
• The subject identifies who is doing the action. An example is the hen in imprinting. This identification is important both in perception and for memory recall.
• The verb identifies the action, such as walking (for imprinting).
• The object may identify a target or some other context. A kernel of food may be such a target for a chick.

This format for information representation supports the conversion from perception to action which is required for imitation

• Alternative representations that support the conversion from vision to action are not readily available
• Up to this point, the author has not been able to think of a reasonable alternative

The conversion from description to command requires a concept of future. The description is seen as specifying a future state of the imitator, and the commands are methods to bring about that future state.

• Since we are claiming that chicks can do such imitation in imprinting, we are forced to conclude that they can handle at least a short term and limited projection of the future.
• This issue will be explored in more detail in the next chapter.

We then have to show that this format is compatible with the information processing functions discussed earlier, such as simple learning.

• The method must be shown as effective for compression, thus extending memory.