Feature selection and classification: A developmental study

Two hundred and thirteen children between the ages of 4 and 16 sorted 30 complex objects freely into groups. These were either schematic faces, schematic imaginary animals, or histoform patterns and were constructed from 9 independently and continuously varying features, each with 10 levels. The three sets were equivalent in terms of underlying numerical structure. A new method of analysis of the groups formed yielded the following direct measures of subjective organization in the sorting of each child: the number of features used in sorting, the accuracy of sorting, and the number of independent dimensions of classification. Both the material and the analysis of sorting were devised on the assumption that only some of the component features of complex objects are used for classification, while others are ignored. Sorting performance differed markedly both with age and with different types of material. The improvement in sorting ability with age, except for the youngest children, could be attributed to an increase in the number of features used rather than an improvement in the accuracy of use. In contrast, the poor sorting of histoform patterns as opposed to faces and animals was due to a low level of sorting accuracy. The basic results on the development of classification confirmed previous studies, and hence suggest that it is valid to consider feature selection as a basic component of classification.     

Blocking of a taste aversion by prior pairings of exteroceptive stimuli with illness

Although rats preferentially associate gustatory stimuli with illness, this does not imply that other types of stimuli cannot be associated with illness. In Experiment 1, rats consuming tap water in a distinctive environment prior to illness subsequently suppressed ingestion in that environment, whereas rats poisoned or injected with saline 10 hr later did not. It was also found that rats poisoned after drinking saccharin in the environment paired with illness developed weaker saccharin aversions than identically trained rats which drank saccharin in a different location. In Experiment 2, degrading the correlation between injection-related cues and illness by interpolating saline injections between LiCl preexposures attenuated the decremental effects of illness preexposure. The strength of the exteroceptive cue-illness associations observed in these studies suggests that theories based on evolutionary arguments are not the most appropriate frame of reference for analyzing stimulus selection in poison-induced avoidance.     

Hemispheric differences in processing emotions and faces.

Visual field differences for the recognition of emotional expression were investigated using a tachistoscopic procedure. Cartoon line drawings of five adult male characters, each with five emotional expressions ranging from extremely positive to extremely negative, were used as stimuli. Single stimuli were presented unilaterally for 85 msec. Subjects (N = 20) were asked to compare this target face to a subsequent centrally presented face and to decide whether the emotional expressions of the two faces, or the character represented by the two faces, were the same or different. Significant left visual field (LVF) superiorities for both character and emotional expression recognition were found. Subsequent analyses demonstrated the independence of these effects. The LVF superiority for emotional judgments was related to the degree of affective expression, but that for character recognition was not. The results of this experiment are consistent with experimental and clinical literature which has indicated a right hemispheric superiority for face recognition and for processing emotional stimuli. The asymmetry for emotion recognition is interpreted as being an expression of the right hemisphere's synthetic and integrative characteristics, its holistic nature, and its use of imagic associations.     

Hesitation and the production of verbal paraphasias and neologisms in jargon aphasia.

Techniques of hesitation analysis taken from studies of normal speakers were applied to the speech of a jargon aphasic. Neologisms were found to follow pauses indicating a word-finding difficulty. Other language functions—phonology, morphology, and syntax—appeared unimpaired, and further analyses of the linguistic and temporal characteristics indicated a single functional disorder in which there is a failure in the mechanisms which associate word-sounds with word-meanings. The patient strategically adapts to this functional impairment by substituting a neologism when lexical search fails. The source of a large class of neologisms, it is hypothesized, is a device which quasirandomly combines English phonemes in a phonotactically regular way. The implications for recovery patterns in jargon aphasia are discussed; and the implications of this case for models of normal language production are explored.     

The effects of symmetry and contour on recognition memory in children

Children in prekindergarten, kindergarten, and second grade were tested on a delayed match-to-sample task using abstract visual patterns. The patterns varied both in type of visual organization (unstructured, diagonally symmetrical, vertically symmetrical, and horizontally symmetrical) and in amount of contour. Following the initial memory task, we attempted to train half of the children at each grade by directing their attention to axes of symmetry; the other half of the children received a control task. Subsequently, the subjects were given a second delayed match-to-sample test. On both pre- and post-training trials, structure influenced performance, especially in the two younger groups. Vertical and horizontal symmetry generally facilitated performance in the prekindergarteners, while all three types of symmetry facilitated performance in the kindergarteners. In addition, children generally responded more accurately to patterns with lower levels of contour. They also made more like-contour than unlike-contour confusions, indicating that quantitative aspects of patterns were encoded. Group differences suggested that both processing capacity and memory increase during the age range studied. Finally, there was no indication that training improved performance at any age.     

The role of exploration in win-shift and win-stay performance on a radial maze

Win-shift spatial memory tasks in a radial maze reinforce animals for avoiding previously visited rewarded arms; win-stay tasks reinforce them for returning to those arms. Win-shift tasks have generally been found much easier to perform, and this may be explained either in terms of foraging models which postulate avoidance of locations where food has been found, or in terms of the predominance of spontaneous alternation (exploration). Experiment 1 examined spontaneous alternation behavior in the radial maze as a function of whether the first visit to an arm had been rewarded or not, and showed that alternation was more probable after nonreward than after reward in both hungry and thirsty rats (a result which conflicts with the foraging account of the win-shift superiority). Experiment 2 replicated the finding that win-stay discrimination performance was inferior to win-shift. A manipulation (lengthening the delay between initial and test choices) which weakens spontaneous alternation, reduced, but did not reverse, the win-shift superiority. In Experiment 3, in order to eliminate the influence of spontaneous alternation, versions of the win-stay and win-shift tasks were devised in which, unlike the original task, all arms were familiar at the choice trial. Under those conditions win-stay was performed better than win-shift. It is concluded that spontaneous alternation plays a major role in many spatial memory tasks, and that the results can best be accounted for by combining principles of exploration and simple associative learning, without recourse to foraging models.     

Quantitative CT scan studies in aphasia. I. Infarct size and CT numbers

Infarct si?e (number of 1-mm² pixels in the lesion) on CT scans of 30 aphasia patients was obtained with a semiautomated computer program. The mean number of lesion pixels present per slice containing lesion was approximately 500 for mild aphasias (transcortical motor and conduction), 700 for Wernicke's, 1000 for Broca's, 1500 for mixed, and 2000 for globals. These differences were significant for $\text{11}\text{15}$ of the group pairwise comparisons. When lesion locus was controlled for in the anterior/posterior plane, 73–100% of the aphasia patients were correctly classified as to type of aphasia by a discriminant analysis utilizing only the number of lesion pixels present on two CT slices. Different slice combinations were used for different aphasia group comparisons. There was a significant correlation between severity of aphasia and lesion size. There was a significant correlation between lesion size and the CT numbers in the lesion. This type of analysis may be useful to predict the prognosis for recovery potential in aphasics who have CT scans performed at 2 months poststroke.