Age-related striatal BOLD changes without changes in behavioral loss aversion

Loss aversion (LA), the idea that negative valuations have a higher psychological impact than positive ones, is considered an important variable in consumer research. The literature on aging and behavior suggests older individuals may show more LA, although it is not clear if this is an effect of aging in general (as in the continuum from age 20 and 50 years), or of the state of older age (e.g., past age 65 years). We also have not yet identified the potential biological effects of aging on the neural processing of LA. In the current study we used a cohort of subjects with a 30 year range of ages, and performed whole brain functional MRI (fMRI) to examine the ventral striatum/nucleus accumbens (VS/NAc) response during a passive viewing of affective faces with model-based fMRI analysis incorporating behavioral data from a validated approach/avoidance task with the same stimuli. Our a priori focus on the VS/NAc was based on (1) the VS/NAc being a central region for reward/aversion processing; (2) its activation to both positive and negative stimuli; (3) its reported involvement with tracking LA. LA from approach/avoidance to affective faces showed excellent fidelity to published measures of LA. Imaging results were then compared to the behavioral measure of LA using the same affective faces. Although there was no relationship between age and LA, we observed increasing neural differential sensitivity (NDS) of the VS/NAc to avoidance responses (negative valuations) relative to approach responses (positive valuations) with increasing age. These findings suggest that a central region for reward/aversion processing changes with age, and may require more activation to produce the same LA behavior as in younger individuals, consistent with the idea of neural efficiency observed with high IQ individuals showing less brain activation to complete the same task.

Discriminative training of the hidden vector state model for semantic parsing

In this paper, we discuss how discriminative training can be applied to the hidden vector state (HVS) model in different task domains. The HVS model is a discrete hidden Markov model (HMM) in which each HMM state represents the state of a push-down automaton with a finite stack size. In previous applications, maximum-likelihood estimation (MLE) is used to derive the parameters of the HVS model. However, MLE makes a number of assumptions and unfortunately some of these assumptions do not hold. Discriminative training, without making such assumptions, can improve the performance of the HVS model by discriminating the correct hypothesis from the competing hypotheses. Experiments have been conducted in two domains: the travel domain for the semantic parsing task using the DARPA Communicator data and the Air Travel Information Services (ATIS) data and the bioinformatics domain for the information extraction task using the GENIA corpus. The results demonstrate modest improvements of the performance of the HVS model using discriminative training. In the travel domain, discriminative training of the HVS model gives a relative error reduction rate of 31 percent in F-measure when compared with MLE on the DARPA Communicator data and 9 percent on the ATIS data. In the bioinformatics domain, a relative error reduction rate of 4 percent in F-measure is achieved on the GENIA corpus.