Recent research has shown that similarity plays crucial roles not only in lower-level cognition, but also in higher-level ones such as problem-solving and learning including analogy. At the same time, these findings have made clear the insufficiency of the traditional model of similarity that assumes a flat representation of a fixed set of properties of an object. This paper explores the possibility of extending similarity to the variety of cognitive activities. For this purpose, by reviewing recent studies on similairty judgment, we examined whether human similarity judgment could reflect a.) structural information, b.) task goals, and c.) knolwedge relevant to achieving the task. The review revealed that human mechanisms of similarity judgments incorporate structural and goal-related information in computing similarity.

抄録全体を表示

Similarity and categorization are inextricably linked in cognitive theory. Research on similarity suggests that comparisons involve a process of structural alignment that compares pairs of structured relational representations. Structural alignment is derived from structure-mapping theory, which was originally designed to account for people's ability to process analogies. The structural alignment process focuses people on common relations that hold between the items being compared. This process yields the commonalities of the pair as well as two kinds of differences: alignable differences, which are directly related to the commonalities, and nonalignable differences, which are unrelated to the commonalities. Evidence for this process has been obtained using studies of comparisons of people's existing categories. These data suggest that category representations are structured so that contrasting categories are easy to compare, but more distant categories are difficult to compare. This pattern of data raises the question of how categories come to be structured such that pairs of similar categories are easier to compare than are pairs of dissimilar categories. A possibility explored here is that the category representations are constructed by a reminding-based process that uses elements of representations of categories that are already known.

抄録全体を表示

Similarity depends on representations of stimuli that are constructed and changed during comparison-making. Specific features may be selectively weighted during comparison, and the features used in a comparison may themselves be a product of the comparison process. Traditional models of similarity and analogy rely on representations that are assumed to exist prior to comparison and are inflexible. Evidence from previous research indicates that weighting of features in similarity judgments may vary dynamically during processing (Goldstone, 1994; Goldstone & Medin, 1994). SIAM (Goldstone, 1994), a model providing an account of dynamic weighting, is discussed. Additional studies indicate that features may be developed or introduced during similarity judgments. A methodology for examining process-oriented models that may account for flexible representations is proposed.

抄録全体を表示

This paper presents a broad overview of the Metacat project, an extension of the Copycat computer model of fluid concepts, high-level perception, and analogy-making. Copycat models the complex, subconscious interplay between concepts and perception that gives rise to the flexible human ability to perceive apparently-dissimilar situations as being “the same”. A key feature of the architecture is the emergence of statistically-robust, high-level behavior from the interactions of many small, low-level, nondeterministic processing agents. All processing occurs through the collective actions of many agents working in parallel on different aspects of an analogy problem, without any higher-level executive process controlling the course of events. Current work on Metacat is focused on extending the Copycat model in a way that permits it to create much richer representations of the analogies it makes. This involves incorporating a long-term memory into the architecture, along with a “self-watching” ability, so that the program can recognize, remember, and recall important patterns that occur in its own processing as it solves analogy problems. Using this higher-order “meta-level” information, analogies can be compared and contrasted in an insightful way, allowing Metacat to understand and explain its answers in a way that Copycat cannot. Metacat's relation-ship to other work in AI and cognitive science is also examined, in particular work on case-based reasoning and derivational analogy.

抄録全体を表示

Two experiments investigated the role of abstraction of the cue expression, which may influence the reminding of the stories. Three kinds of abstraction were used for cue expression, 1) Lower level of abstraction: concrete description of the event according to the temporal sequence, 2) Higher level of abstraction: short description with a few abstract words (e.g., “It's Too Much Late”), 3) Moderate level of abstraction: short description of the event with the thematic relation like proverb (e.g., “Closing the Barn Door After the Horse Is Gone”). The results of the experiments indicated that the pattern of reminding is influenced by the levels of abstraction of the cue expression, and that the cue expression with moderate level of abstraction like proverb is the more effective than the others to remind the story with same thematic relation. Discussion focuses on the role of the description of the typical case to refer to the thematic relation.

抄録全体を表示

Based on the diagnosticity principle (Tversky, 1977), the systematicity principle (Gentner, 1983) is interpreted as that the deep structure of object has higher diagnostic value than the surface features in analogical reasoning. However, it is not clear that which one of the object's features corresponds to each of the surface and the deep structure in Tversky's research. Another question is whether the diagnostic value of the deep structure is judged higher than that of the surface or not when the diagnostic value of the object's feature is determined by a presentational factor such as the combination of base analogues. In the present paper, we conducted 2 experiments to analyze the effect of presentational factor on the diagnostic value of a given feature using the competitive design in the context of the diagnosticity principle.
In experiment 1, 82 subjects read only one judicial precedent and composed a sentence of the target trial. The results supported the validity of the systematicity principle except the subjects' plausibility judgements for their sentence predictions. In experiment 2, 60 subjects read 3 judicial precedents and judged the similarity with the target trial to compose a sentence. A diagnosticity effect was found in the subjects' similarity judgments and also in their importance judgements of the target trial's superficial features. The subjects judged the base-to-target similarity higher with the structurally dissimilar base than the structurally similar base, and their sentence predictions did not match the sentence of the original trial. The results did not fully support the validity of the systematicity principle.

抄録全体を表示

This paper proposes implicit display theory of verbal irony that overcomes several difficulties of previous irony theories. The theory claims that irony implicitly displays the fact that its utterance situation is surrounded by ironic environment consisting of three properties, but hearers recognize an utterance to be potentially ironic even when they do not see all the three properties implicitly displayed by the utterance. Implicit display of ironic environment is accomplished in such a way that an utterance alludes to the speaker's expectation, violates one of pragmatic principles, and is accompanied by several cues for implying the speaker's negative emotional attitude. This paper also shows the validity of implicit display theory by explaining how the theory is consistent with various aspects of irony such as its echoic nature, ironic cues, unintentional irony, the asymmetry of irony and so on.

抄録全体を表示