"... In PAGE 1: ... Finally, the last section discusses the generalizability of the HAROLD model in terms of cognition, the brain, and the elderly population. Evidence of Age-Related Asymmetry Reductions Table1 summarizes functional neuroimaging evidence consis- tent with the HAROLD model in the domains of episodic memory retrieval, episodic memory encoding/semantic memory retrieval, working memory, perception, and inhibitory control. Each of the first five subsections below reviews the findings in one of these domains.... In PAGE 3: ... In younger adults, left PFC activity was nearly twice as large as was right PFC activity. In older adults, left PFC activity was reduced but right PFC was not and, as a result, the asymmetry shown by younger adults was eliminated (see Table1 ). According to the authors, the age-associated reduction in left prefrontal activation eliminated the left hemisphere asymmetry evident in the younger participants, who exhibited twice as much activation in left relative to right prefrontal activation (p.... In PAGE 3: ...ith a semantic processing strategy (i.e., environmental support) increased their PFC activity in both hemispheres but did not increase their hemispheric asymmetry (see Table 1). As illustrated in Table1 , the results in the episodic encoding/ semantic retrieval domain support the generalizability of the HAROLD model in two ways. First, they demonstrate that age- related asymmetry reductions are not limited to the situation in which PFC activity is right lateralized in young adults with an age-related increase in left PFC activity; they may also occur when PFC activity is left lateralized in young adults with an age-related decrease in left PFC activity.... In PAGE 4: ... The fact that an age-related asymmetry reduction was found in middle-aged individuals suggests that the HAROLD pattern develops before old age. As shown in Table1 , age-related asymmetry reductions can be found when PFC activity is right lateralized in young adults, as in the case of episodic retrieval, as well as when PFC activity is left lateralized in young adults, as in the case of episodic encoding/ semantic retrieval. The results in the working memory domain indicate that this is also true when the task is constant and that it is the nature of the processed information that affects the lateral- ization of PFC.... In PAGE 4: ... An fMRI study that investigated this paradigm (Garavan, Ross, amp; Stein, 1999) associated response inhibition with a network of regions that are strongly lateralized to the right hemisphere, which includes PFC and parietal regions. Thus, in young adults, inhibitory control was associated with right PFC activity (see Table1 ). Nielson, Langenecker, amp; Garavan (2002) investigated the same paradigm in a group of older adults.... In PAGE 4: ... Nielson, Langenecker, amp; Garavan (2002) investigated the same paradigm in a group of older adults. In older adults, inhibitory control elicited significant activity not only in right PFC but also in left PFC (see Table1 ). As noted later, age-related increases in the left hemisphere were also found in the parietal cortex, suggesting that the HAROLD model may gener- alize beyond PFC.... In PAGE 5: ... Finally, in the case of perception and inhibitory control, the age-related asymmetry reduction primarily involved increases in left PFC. Thus, the HAROLD model is consistent with a variety of age-related changes in activity (see two rightmost columns in Table1 ). As the example in Figure 3 illustrates, this diversity may be a consequence of the use of brain images that display differ- ences in activations only if they exceed a certain significance threshold.... In PAGE 5: ...l., 1998; Nagahama et al., 1997; Rypma, Prabhakaran, Desmond, amp; Gabrieli, 2001) but are not listed as evidence for the HAROLD model because they did not explicitly report an age-related asym- metry reduction. Therefore, age-related asymmetry reductions are easier to detect in Outcomes D and E, and this explains why Table1 consists mainly of these two types of outcomes. However, age-related asymmetry reductions could be responsible for Out- comes A, B, and C in many studies but were simply not detected.... In PAGE 11: ... As the neural correlates of attentional resources, inhibitory control, and process- ing speed are differentiated, so will be the predictions that the resources, inhibition, and speed views of aging make about the HAROLD model. Generalizability of Age-Related Asymmetry Reductions Generalizability Within Cognition Although the evidence listed in Table1 indicates that age- related asymmetry reductions occur for a variety of different cognitive functions, there are several unresolved issues regarding the generalizability of the HAROLD model within cognition. For example, it is unclear if age-related asymmetry reductions occur only for higher order cognitive processes, such as the ones listed in Table 1, or whether they can also be found for simple sensory and motor processes.... In PAGE 11: ... Generalizability of Age-Related Asymmetry Reductions Generalizability Within Cognition Although the evidence listed in Table 1 indicates that age- related asymmetry reductions occur for a variety of different cognitive functions, there are several unresolved issues regarding the generalizability of the HAROLD model within cognition. For example, it is unclear if age-related asymmetry reductions occur only for higher order cognitive processes, such as the ones listed in Table1 , or whether they can also be found for simple sensory and motor processes. The answer to this question is informative with respect to psychogenic versus neurogenic accounts of the HAROLD model.... ..."

"... In PAGE 4: ... The scenario described above indicates that a proper IT infrastructure is fundamental to accomplish fMRI studies successfully. Table1 summarizes the different problems faced for data management in fMRI studies and the challenges posed to the construction of an adequate infrastructure. 4.... In PAGE 5: ... This application therefore requires coordinated resource sharing and problem solving in dynamic, multi-institutional virtual organizations (VOs), which is the goal of grid technologies [6]. A virtual laboratory for fMRI (VL-f) is under development in the scope of the VL-e project to address some of the challenges listed in Table1 . The goal is to construct a shared computational infrastructure with hardware, software and services to efficiently, reliably and securely perform (large scale) fMRI studies.... ..."

"... In PAGE 3: ... Materials Two sets of word/nonword pairs, gift--kift and giss--kiss, were selected that met the following criteria: they were one-syllable [StopVC(C)] stimuli that shared the same vowel, and the same manner of articulation in final consonant(s); to the extent possible, word endpoints were matched for frequency (Kucera and Francis 1967), and word and nonword members of each pair were matched for both neighborhood density (Luce and Pisoni 1998), and phonotactic probability (Vitevitch et al. 1999) (see Table1 for a list of the lexical properties of these stimuli). In order to establish a neutral baseline against which lexically shifted identification functions could be compared, a nonword to nonword continuum, gish--kish, was also chosen, matched as close as possible to the word/nonword pairs in terms of density and phonotactic probability.... ..."

"... In PAGE 4: ... 2. Estimated visual task-related time courses and the corresponding activation maps of Infomax (a, b) and Info- max with feature-selective projection (c, d) The correlation between the experimental paradigm and the estimated visual task-related time course, rt^ t, is com- puted and listed in Table3 . Figure 2 shows the estimated time courses and the associated activation maps (Z-scored with Z gt; 2:5).... ..."

"... In PAGE 7: ... We re- peated the process of training and testing for 10 times and report the average accu- racy. The first row of Table2... In PAGE 8: ...discriminative features selected, as well as subsets of them based on the significance weights, wk, obtained by the proposed approach. The second row of Table2 shows the comparative results when using the forward feature selection (FFS) approach. In all the experiments we used the initial characterization attribute values.... ..."

"... In PAGE 7: ... We re- peated the process of training and testing for 10 times and report the average accu- racy. The first row of Table2 includes the classification results when using all the ... In PAGE 8: ...weights, wk, obtained by the proposed approach. The second row of Table2 shows the comparative results when using the forward feature selection (FFS) approach. In all the experiments we used the initial characterization attribute values.... ..."