... methodology.¹

The earlier web version of this paper mistakenly ascribed the blank slate doctrine to Descartes.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... prejudices.²

There is a complication. Appropriate experience is often required for the genetically determined structures to develop properly, and much of this experience can be regarded as learning.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... prejudices?³

I don't argue that a Lockean baby wouldn't work at all. Only that it would have a much longer babyhood than human babies do. Even ``universal grammar'' might be learned from experience. I just think evolution has learned to build in many of these features. Therefore, it is an empirical question whether a particular ability is learned or innate.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... actions.⁴

We emphasize the effect of the actions on the world and not the new sensations that result from the action.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... stance.⁵

Designer stance is related to Daniel Dennett's design stance [Den78], but Aaron Sloman has persuaded me that I was not using it quite in the way Dennett used design stance. 1999 note: Dennett has approved this usage of design stance, but I still want a distinct term.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... world.⁶

Whether a baseball will pass over the plate is approximately deterministic once it leaves the pitcher's hand, but the batter and the spectators have to guess.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... innate.⁷

Kant distinguished between appearance and reality, but AI and psychology need to study the distinction at a more mundane level than Kantian philosophers have brought themselves to do. I don't believe the study of platonic or neo-platonic forms will help understand the relation between physical dogs and the various ways they could affect the senses of a child or robot.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... kind.⁸

It might be more parsimonious intellectually to have just a relation of similarity between objects. However, the world as it is justifies the bolder attitude that there are kinds, and we should build this into our robots and expect it in our children. The use of nouns in language presupposes more than just similarity relations.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... reptiles.⁹

Here the facts of evolution have an observable payoff akin to Haeckel's ``Ontogeny recapitulates phylogeny''.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... universal.¹⁰

This may be worth a small pound on the table. It would seem that a person, and maybe even some animals, can test whether a goal is achievable by parsing the goal regression structure. Of course, there are limits in how big a structure can be parsed, but the competence puts no limit on the size. It may be that goal-regression memory is in addition to other short term memory and can only be used in connection with remembering goals.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... places.¹¹

Astronomers use ``principle of mediocrity'' for the hypothesis that there is nothing special on the average about our own part of the universe or about our own point in time.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... language¹²

Logicians do not consider logic itself to be a language, but rather consider a language to be defined by the predicate and function symbols that are used with the logic. This is a valuable distinction and AI and cognitive science researchers should maintain it.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... cortex.¹³

A robot might tell another robot, ``Look through my eyes, and you'll see it.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... logic.¹⁴

There also has to be a program using the logical sentences, and efficiency will very likely require it to use declaratively expressed heuristics to guide its search. Very little progress has been made in that direction, so we will ignore heuristic control in this article.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... height.¹⁵

Someone tried this on the child of a prominent AI researcher, eliciting the answer, ``Oh, I'm not old enough to have conservation yet''.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

....¹⁶

Conservation of heat wasn't apparent to 17th century Italian experimenters such as Toricelli who used a flow model of heat. It wasn't until 1750 that Thomas Black discovered that heat could be regarded as a quantity that moved from one object to another by conduction.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... conservation.¹⁷

Archimedes assumed that the volume of the king's crown was equal to the volume of the water it would displace, so he didn't need to make detailed measurements on the crown.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... way.¹⁸

Some aspects of computing are going from serial to parallel in order to achieve greater speed, but a lot of communication within and among computers is going the other way for greater simplicity and reliability. Multi-wire cables are being replaced by single Ethernet or fiber optic cables. Truth and beauty are not to be found in a single direction.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.