Probabilistic category learning engages neural circuitry that includes the prefrontal cortex and caudate nucleus, two regions that show prominent changes with normal aging. However, the specific contributions of these brain regions are uncertain, and the effects of normal aging have not been examined previously in probabilistic category learning. In the present study, using a blood oxygenation leveldependent functional magnetic resonance imaging block design, 18 healthy young adults (mean age, 25.5 Ϯ 2.6 years) and 15 older adults (mean age, 67.1 Ϯ 5.3 years) were assessed on the probabilistic category learning "weather prediction" test. Whole-brain functional images acquired using a 1.5T scanner (General Electric, Milwaukee, WI) with gradient echo, echo planar imaging (3/1 mm; repetition time, 3000 ms; echo time, 50 ms) were analyzed using second-level random-effects procedures [SPM99 (Statistical Parametric Mapping)]. Young and older adults displayed equivalent probabilistic category learning curves, used similar strategies, and activated analogous neural networks, including the prefrontal and parietal cortices and the caudate nucleus. However, the extent of caudate and prefrontal activation was less and parietal activation was greater in older participants. The percentage correct and reaction time were mainly positively correlated with caudate and prefrontal activation in young individuals but positively correlated with prefrontal and parietal cortices in older individuals. Differential activation within a circumscribed neural network in the context of equivalent learning suggests that some brain regions, such as the parietal cortices, may provide a compensatory mechanism for healthy older adults in the context of deficient prefrontal cortex and caudate nuclei responses.

No abstract

The performance of a welding process determines not only the cost, but also the quality of the product. How to control the welding process in order to ensure good welding performance with less cost and higher Productivity has become critical. The objective of this study is twofold: (1) developing artificial neural networks to predict welding performance using different learning algorithms: back propagation, simulated annealing and tabu search; (2) comparing and discussing the performance of neural networks trained using those algorithms. Statistical analysis shows that back propagation is able to make more accurate prediction than the other algorithms for this particular application. However, all three algorithms demonstrate impressive flexibility and robustness.