How Should a Speech Recognizer Work?

Although researchers studying human speech recognition (HSR) and automatic speech recognition (ASR) share a common interest in how information processing systems (human or machine) recognize spoken language, there is little communication between the two disciplines. We suggest that this lack of communication follows largely from the fact that research in these related fields has focused on the mechanics of how speech can be recognized. In Marr's (1982) terms, emphasis has been on the algorithmic and implementational levels rather than on the computational level. In this article, we provide a computational-level analysis of the task of speech recognition, which reveals the close parallels between research concerned with HSR and ASR. We illustrate this relation by presenting a new computational model of human spoken-word recognition, built using techniques from the field of ASR that, in contrast to current existing models of HSR, recognizes words from real speech input.     

A Meta-Analysis of Cognitive Deficits in Adults with a Diagnosis of Schizophrenia

This review identified 1275 studies examining cognitive deficits in people with schizophrenia, published between 1990 and 2003. Data from 113 studies (4365 patients and 3429 controls) were combined in a meta-analysis carried out on the five cognitive domains of IQ, memory, language, executive function, and attention. Studies were excluded where they lacked a suitable control group or failed to present complete information. In all five cognitive domains, analysis indicated a consistent trend for patients to perform more poorly than healthy controls, with significant heterogeneity across studies. Sources of heterogeneity were analyzed and a need to ensure more appropriate composition of patient and control groups and to adopt a more refined and methodologically correct, hypothesis-driven approach was identified.     

The visual search analogue of latent inhibition: Implications for theories of irrelevant stimulus processing in normal and schizophrenic groups

Latent inhibition (LI) is a robust phenomenon that is demonstrated when a previously inconsequential stimulus is less effective in a new learning situation than a novel stimulus. Despite LI’s simplicity, there is considerable disagreement as to its theoretical basis. Attentional theories claim that unattended stimulus preexposures reduce stimulus associability. Alternatively, it has been asserted that associability is unaffected and that LI is a result of competition/retrieval processes. The present article reviews a series of visual search studies, some with normal subjects, both undifferentiated and divided into low and high schizotypals, and others with pathologies that entail dysfunctional attention, such as schizophrenia, Parkinson’s disease, and anxiety. The visual search conditions were designed to model those of traditional LI experiments, while tapping attentional processes independently of the learning scores that index LI. A variety of evidence from these and other studies is used to support the involvement of attentionaland retrieval processes in LI. A model of the mechanism of action of these processes in LI is presented, together with its application to schizophrenia.     

Bayesian eggs and Bayesian omelettes: reply to Stern (2005)

In this response to Stern's (2005) discussion of Klugkist, Laudy, and Hoijtink (2005), model inference based on posterior probabilities on the parameter space is discussed. Furthermore, the authors respond to Stern's example in which all possible orderings are included via a short discussion of exploratory versus theory-based modeling. Finally, the authors show that the Bayesian approach is flexible and can deal with many types of constraints. This is illustrated using a model with constraints on the differences between means.     

Processing demands in belief‐desire reasoning: inhibition or general difficulty?

Most 4-year-olds can predict the behavior of a person who wants an object but is mistaken about its location. More difficult is predicting behavior when the person is mistaken about location and wants to avoid the object. We tested between two explanations for children's difficulties with avoidance false belief: the Selection Processing model of inhibitory processing and a General Difficulty account. Children were presented with a false belief task and a control task, in which belief attribution was as difficult as in the false belief task. Predicting behavior in light of the character's desire to avoid the object added more difficulty in the false belief task. This finding is consistent with the Selection Processing model, but not with the General Difficulty account.     

Core mechanisms in "theory of mind"

Our ability to understand the thoughts and feelings of other people does not initially develop as a theory but as a mechanism. The "theory of mind" mechanism (ToMM) is part of the core architecture of the human brain, and is specialized for learning about mental states. Impaired development of this mechanism can have drastic effects on social learning, seen most strikingly in the autistic spectrum disorders. ToMM kick-starts belief-desire attribution but effective reasoning about belief contents depends on a process of selection by inhibition. This selection process (SP) develops slowly through the preschool period and well beyond. By modeling the ToMM-SP as mechanisms of selective attention, we have uncovered new empirical phenomena. We propose that early "theory of mind" is a modular-heuristic process of domain-specific learning.     

Rejection sensitivity and the defensive motivational system: insights from the startle response to rejection cues

Rejection sensitivity (RS) is the disposition to anxiously expect, readily perceive, and intensely react to rejection. This study used the startle probe paradigm to test whether the affect-based defensive motivational system is automatically activated by rejection cues in people who are high in RS. Stimuli were representational paintings depicting rejection (by Hopper) and acceptance (by Renoir), as well as nonrepresentational paintings of either negative or positive valence (by Rothko and Miro, respectively). Eyeblink startle magnitude was potentiated in people high in RS when they viewed rejection themes, compared with when they viewed nonrepresentational negative themes. Startle magnitude was not attenuated during viewing of acceptance themes in comparison with nonrepresentational positive themes. Overall, the results provide evidence that for people high in RS, rejection cues automatically activate the defensive motivational system, but acceptance cues do not automatically activate the appetitive motivational system.     

Postextinction infusion of a mitogen-activated protein kinase inhibitor into the medial prefrontal cortex impairs memory of the extinction of conditioned fear

We investigated whether postextinction training infusion of PD098059, a selective inhibitor of mitogen-activated protein kinase (MAPK) activation, into the medial prefrontal cortex, would impair retention of extinction learning in rats. We found that immediate, but not late (2 or 4 h), postextinction infusion of PD098059 provoked a full return of conditioned freezing. These results suggest that activation of prefrontal MAPK in early stages of postextinction training participates in processes that protect against spontaneous recovery of aversive responses.     

How the mind hurts and heals the body

The author reviews some of the social and behavioral factors acting on the brain that influence health, illness, and death. Supported with data from several areas of research, his proposal for understanding health and illness provides both the concepts and the mechanisms for studying and explaining mind-body relationships. The brain is the body's first line of defense against illness, and the mind is the emergent functioning of the brain. This mind-body approach incorporates ideas, belief systems, and hopes as well as biochemistry, physiology, and anatomy. Changing thoughts imply a changing brain and thus a changing biology and body. Belief systems provide a baseline for the functioning brain upon which other variables act and have their effects.