The Universal Theory Model of Concepts and the Dissolution of the Puzzle of Concept Acquisition

I present a Universal Theory Model of Concepts (UTMC) that helps dissolve a well-known Puzzle of Concept Acquisition: how can a person ever acquire a “new” concept? The key state variables of the UTMC are illustrated in an microgenesis experiment where adult subjects attempt to learn the meaning of 3 new verbs in a “Causal Blocksworld” computer application. The puzzle is dissolved by recognizing the 2 viewpoints that one can take on the UTMC. The Microgenesis of New Concepts In this paper I help dissolve a long-standing dispute between the nativist Jerry Fodor and the developmentalist concerning a fundamental Puzzle of Concept Acquisition: (Fodor 1975, Laurence and Margolis 2002) How can a person acquire a genuinely new concept? I dissolve the puzzle by describing an experiment concerning the microgenesis of a set of “new” concepts that adults acquire in a single experimental setting. To make sense of the microgenesis (or lack thereof), a new model of concepts is required. I call it the Universal Theory Model of Concepts (UTMC). I believe this model is latent in the dispute, and that it is possible to lay the dispute to rest by showing that the source of the dispute is due to the choice of viewpoint enabled in this model. In this experiment, subjects try to discover the 3 laws that govern the behavior of 29 blocks within a computer application called “Causal Blocksworld”. Ten adult subjects participated in this experiment. Unknown the subjects at the beginning of the experiment, there are four kinds of blocks: As, Bs, Cs and Ds. All four kinds of blocks are perceptually indistinguishable except an alphabetic label; for each of the 4 kinds, a block is the kind of block it is because of what it can activate, and what can activate it. Three laws govern block activation: • lawab: When an block of kind A touches a block of kind B, the B block lights up • lawc: When a block of kind C block touches another block of kind C, the one with a lower “power” α will light up, where each C block’s power α is unchangable. • lawd: When a block of kind D touches another block of kind D, one of them will light up (or not), consistently, across different activations. 29 blocks are introduced to the subject in a staging area in 5 phases (9 in phase 1 and then 5 for each phase thereafter). Subjects engage in free-form play with the objects to discover the above laws, and end up organizing the blocks spatially into clusters, shown in Figure 1(a) for 3 kinds of subjects. Moreover, subjects are also given cues to the meaning of 3 new verbs – gorp, pilk, and seb in a Word Cue area (see Figure 1(a)) for a subset (onethird) of the blocks: • lawab: When an A block labeled x activates a B block labeled y, subjects are shown “x is gorping y” • lawc: When a C block labeled x activates a C block labeled y, subjects are shown “x is pilking y” • lawd: When a D block labeled x activates a D block labeled y, subjects are shown “x is sebbing y” In each of the 5 phases, subjects are asked questions to test whether they have learned the meaning of the 3 verbs. I tested them with two methods: getting their plain text definitions of the 3 verbs and giving them a forced-choice naming condition test for each of the 3 verbs. Plain text definitions were requested in the following way. At the end of phase 1 and at the end of phase 5, subjects are asked 1 question of each of the 3 verbs within the application: “What do you think (gorp/pilk/seb) means?” and type in plain text responses. After phase 1, subjects uniformly give responses that do not indicate knowledge of more than one kind, saying e.g. “... it means an object lights up another object” or more often “no idea”. But after phase 5, some subjects, but not all, differentiate between the 4 kinds of blocks: • “gorp means to light up one of the group OUYVI” • “seb means for one of the object group in RFIDRPEM to light up another of the same group.” Naming conditions were tested in the following way. After every phase, subjects are required to touch one block (labeled x) to another (labeled y), where one of the blocks lights up. No Word Cue is given. Instead, subjects answer a naming condition question, by choosing which of 6 descriptions best describe what happened: