Do Additional Features Help or Hurt Category Learning? The Curse of Dimensionality in Human Learners

The curse of dimensionality, which has been widely studied in statistics and machine learning, occurs when additional features cause the size of the feature space to grow so quickly that learning classification rules becomes increasingly difficult. How do people overcome the curse of dimensionality when acquiring real‐world categories that have many different features? Here we investigate the possibility that the structure of categories can help. We show that when categories follow a family resemblance structure, people are unaffected by the presence of additional features in learning. However, when categories are based on a single feature, they fall prey to the curse, and having additional irrelevant features hurts performance. We compare and contrast these results to three different computational models to show that a model with limited computational capacity best captures human performance across almost all of the conditions in both experiments.     

Debunking Morality: Lessons from the EAAN Literature

This article explores evolutionary debunking arguments as they arise in metaethics against moral realism and in philosophy of religion against naturalism. Both literatures have independently grappled with the question of which beliefs one may use to respond to a potential defeater. In this article, I show how the literature on the argument against naturalism can help clarify and bring progress to the literature on moral realism with respect to this question. Of note, it will become clear that the objection that the moral realist begs the question, when appealing to the truth of some of her moral beliefs, is unsuccessful.     

On the nature of perceptual limits in vision

Summary Studies of attention have generally assumed that the principal limit to visual perception involves the capacity to identify a stimulus. Analyses of responses in visual tasks, however, suggest that errors can arise from a loss of position information, as well as a loss of identity information, and that these two types of errors may be affected by different variables. Based on this, we have proposed that the difficult aspect of perception is not that of recognizing the form of the stimulus but rather determining its position relative to other items in the field. To demonstrate this, we conducted a series of experiments on lateral masking that showed that the failure to perceive a parafoveal stimulus surrounded by other items is due to two quite separate components: a loss of identity information induced by the immediately adjacent items and a loss of position information induced by both adjacent and non-adjacent items.This research was supported by grant A-9581 from the Natural Sciences and Engineering Research Council of Canada to the senior author. The first two experiments were reported in a paper to the annual meeting of the Psychonomics Society, San Antonio, 1984     

Rearing in response to a visual conditioned stimulus: An index of associative strength or of associability?

Two procedures for observing rearing behavior during a localized visual CS paired with an appetitive US were compared within subjects during classical conditioning and extinction procedures. When rearing was observed using the method and response definition described by P. C. Holland (1977, Journal of Experimental Psychology: Animal Behavior Processes, 3, 77–104), responding was acquired and maintained during conditioning trials in rats receiving paired presentations of the CS and US, but not in rats receiving the stimuli according to a “truly random” procedure. When rearing of the same rats was observed using the method and response definition described by H. Kaye and J. M. Pearce (1984, Journal of Experimental Psychology: Animal Behavior Processes, 10, 90–109), responding declined across conditioning trials in both groups. Furthermore, in extinction, rearing during the CS declined according to Holland's scoring method but was temporarily restored according to Kaye and Pearce's scoring. Observations of a second behavior, magazine responding, revealed essentially the same pattern of responding for both methods of scoring. These results support the position that the response criteria used by Holland and by Kaye and Pearce assess different components of behavior that occur in response to a visual CS, reflecting associative strength (i.e., a CR) and associability of the CS (i.e., an OR), respectively.     

REJOINDER: WHITHER MARITAL THERAPY?

It is apparent that Gurman and Kniskern have greatly revised their original position in response to our critique. We believe that their revised position is more fully consonant with the available empirical evidence, although certain areas of disagreement between us are still evident. Overall, we believe that this dialogue has generated some promising lines for future research and clinical practice.     

Reviews

S tein , M urray . Jung's Treatment of Christianity: The Psychotherapy of a Religious Tradition
R ycroft , C harles . Psychoanalysis and Beyond
C hasseguet -S mirgel , J anine . The Ego Ideal: A Psychoanalytic Essay on the Malady of the Ideal
R othstein , A rnold . The Narcissistic Pursuit of Perfection