Exploring What Is Encoded in Distributional Word Vectors: A Neurobiologically Motivated Analysis

The pervasive use of distributional semantic models or word embeddings for both cognitive modeling and practical application is because of their remarkable ability to represent the meanings of words. However, relatively little effort has been made to explore what types of information are encoded in distributional word vectors. Knowing the internal knowledge embedded in word vectors is important for cognitive modeling using distributional semantic models. Therefore, in this paper, we attempt to identify the knowledge encoded in word vectors by conducting a computational experiment using Binder et al.'s (2016) featural conceptual representations based on neurobiologically motivated attributes. In an experiment, these conceptual vectors are predicted from text-based word vectors using a neural network and linear transformation, and prediction performance is compared among various types of information. The analysis demonstrates that abstract information is generally predicted more accurately by word vectors than perceptual and spatiotemporal information, and specifically, the prediction accuracy of cognitive and social information is higher. Emotional information is also found to be successfully predicted for abstract words. These results indicate that language can be a major source of knowledge about abstract attributes, and they support the recent view that emphasizes the importance of language for abstract concepts. Furthermore, we show that word vectors can capture some types of perceptual and spatiotemporal information about concrete concepts and some relevant word categories. This suggests that language statistics can encode more perceptual knowledge than often expected.     

Depth separation between foreground and background on visually induced perception of self-motion

A uniformly moving visual pattern can induce observer's self-motion perception in the opposite direction (vection), and an additional static stimulus can modulate (facilitate or inhibit) the strength of it. The present study was designed to investigate the effects of stimulus depth order and the depth distances of the visual stimulus on the inhibition and facilitation of vection caused by the additional static stimulus, measuring duration and estimated magnitude of vection as indices of vection strength. Analysis of this psychophysical experiment with four participants indicated that the static foreground presented in front of the moving pattern can facilitate vection, whereas the static background inhibits it (Duration: F1,3= 12.06, p<.05; Estimation: F1,3= 13.87, p<.05). Furthermore, the depth distances from the observer or the depth separation between the foreground and the background did not affect the self-motion perception (F2,6 < 1.0 for duration and estimation).     

Preliminary study of reminiscence therapy on depression and self-esteem in cancer patients

The present study investigated use of life review, a form of reminiscence, on the depression and self-esteem in cancer patients. 15 cancer patients in the experimental group participated in individual reminiscence therapy. 21 patients in the comparison group received no therapy. All patients were measured on both depression and self-esteem scales during two testing periods. Analysis showed mean depression scores of the cancer patients decreased and mean self-esteem increased significandy after the life-review therapy sessions, while the scores of the comparison group did not change. Furthermore, patients' psychological states were improved, and they thought their problems had been addressed. These results suggest reminiscence therapy can be useful for cancer patients.     

The role of language in memory for actions

Languages differ with respect to how aspects of motion events tend to be lexicalized. English typically conflates MOTION with MANNER, but Japanese and Spanish typically do not. We report a set of experiments that assessed the effect of this cross-linguistic difference on participants' decisions in a similarity-judgment task about scenes containing novel animations as stimuli. In Experiment 1, which required participants to encode the stimuli briefly into memory, we observed a language effect; in Experiment 2, which required participants to analyze the same stimuli, but not remember them, the language effect disappeared. Hence, these experiments reveal a task-dependent effect, which, we argue, points to working memory as the source of the language effect observed in Experiment 1 and, potentially, other experiments that have shown a linguistic relativity effect.     

Understanding of the relationship between seeing and knowing by tufted capuchin monkeys (Cebus apella)

The ability of four tufted capuchin monkeys (Cebus apella) to recognize the causal connection between seeing and knowing was investigated. The subjects were trained to follow a suggestion about the location of hidden food provided by a trainer who knew where the food was (the knower) in preference to a trainer who did not (the guesser). The experimenter baited one of three opaque containers behind a cardboard screen so that the subjects could not see which of the containers hid the reward. In experiment 1, the knower appeared first in front of the apparatus and looked into each container; next, the guesser appeared but did not look into any containers. Then the knower touched the correct cup while the guesser touched one of the three randomly. The capuchin monkeys gradually learned to reach toward the cup that the knower suggested. In experiment 2, the subjects adapted to a novel variant of the task, in which the guesser touched but did not look into any of the containers. In experiment 3, the monkeys adapted again when the knower and the guesser appeared in a random order. These results suggest that capuchin monkeys can learn to recognize the relationship between seeing and knowing. Accepted after revision: 10 September 2001 Electronic Publication     

Beyond monoamines: glutamatergic function in mood disorders

The monoamine theory has implicated abnormalities in serotonin and norepinephrine in the pathophysiology of major depression and bipolar illness and contributed greatly to our understanding of mood disorders and their treatment. Nevertheless, some limitations of this model still exist that require researchers and clinicians to seek further explanation and develop novel interventions that reach beyond the confines of the monoaminergic systems. Recent studies have provided strong evidence that glutamate and other amino acid neurotransmitters are involved in the pathophysiology and treatment of mood disorders. Studies employing in vivo magnetic resonance spectroscopy have revealed altered cortical glutamate levels in depressed subjects. Consistent with a model of excessive glutamate-induced excitation in mood disorders, several antiglutamatergic agents, such as riluzole and lamotrigine, have demonstrated potential antidepressant efficacy. Glial cell abnormalities commonly associated with mood disorders may at least partly account for the impairment in glutamate action since glial cells play a primary role in synaptic glutamate removal. A hypothetical model of altered glutamatergic function in mood disorders is proposed in conjunction with potential antidepressant mechanisms of antiglutamatergic agents. Further studies elucidating the role of the glutamatergic system in the pathophysiology of mood and anxiety disorders and studies exploring the efficacy and mechanism of action of antiglutamatergic agents in these disorders, are likely to provide new targets for the development of novel antidepressant agents.     

Visually guided facial actions in rhesus monkeys

We trained 2 monkeys to display facial actions in response to corresponding arbitrary visual cues. Each monkey executed the task successfully, and each displayed two different facial actions corresponding to either hand-sign or color cues. More than 90% of the responses were correct for each monkey. These results provide evidence that monkeys can execute facial actions in response to conditioned visual cues in the absence of social context. These data suggest that facial actions of monkeys are flexible enough for use in further laboratory investigations—for example, in studies on the neural mechanisms underlying the execution of actions.     

Validity of Treatment Target Progress Ratings as Indicators of Youth Improvement

We examined whether clinical progress ratings on the Monthly Treatment and Progress Summary form (MTPS), an idiographic treatment progress measure, were meaningfully related to changes measured by two separate standardized instruments; the Child and Adolescent Functional Assessment Scale (CAFAS) and the Child and Adolescent Level of Care Utilization System (CALOCUS). Validity coefficients were examined at intake and three-, six-, and nine-month follow-up periods. Samples were selected for each measure and follow-up period from the population of youth receiving services through the Hawaii Child and Adolescent Mental Health Division’s (CAMHD) system of care. Significant youth improvement was evident on all three measures across all follow-up intervals. The type of changes measured by the MTPS and CAFAS were more alike over longer follow-up intervals than changes measured by the CALOCUS. The MTPS captures distinct aspects of client change that overlaps somewhat with CAFAS and CALOCUS measures. The MTPS is a brief client-tailored measure that seems to provide valid, sensitive, and nonredundant client specific treatment outcome information that can be collected on a frequent basis within a complex system of care.     

Rorschach Comprehensive System data for a sample of 240 adult nonpatients from Japan

This article presents Rorschach Comprehensive System (CS; Exner, 1995) results for 240 adult nonpatients living in the central area of Tokyo, Japan. Inclusion/exclusion criteria are described. Interrater reliability statistics at the response level are presented along with scores for CS variables.