"... In PAGE 13: ...3 30], van Veen et al., 2001). It included a spherical volume of 33 voxels with a radius of 6 mm. Table5 shows the percent signal change within the ACC in each of the experiments tested. ---------------------- Insert Table 5 here --------------------- The ACC was significantly involved in all but the visual response selection task.... In PAGE 13: ...3 30], van Veen et al., 2001). It included a spherical volume of 33 voxels with a radius of 6 mm. Table 5 shows the percent signal change within the ACC in each of the experiments tested. ---------------------- Insert Table5 here --------------------- The ACC was significantly involved in all but the visual response selection task. On one account, the lack of ACC activation in the visual RS task may be attributed to the blocked design, which involved constant response conflict within a block with correspondingly reduced necessity for conflict monitoring.... ..."

"... In PAGE 14: ...Table 2: Misrouting direction and fault direction determination rule for mes- sages not on the destination column rized in Table 1 and Table2 . Table 1 shows the rules used when the message is in its destination column.... In PAGE 14: ... The rules for the other case is shown in Table 2. Note that rules in Table 1 utilize the message type and fault direction of the message as inputs and rules in Table2 utilize the fault direction and current routing direction of the message as inputs. Our fault tolerant routing algorithm in 2D mesh is summarized in Figure 7.... In PAGE 14: ... At node (2,1), the e-cube hop of M is blocked by a fault and M becomes a misrouting message. By fault direction determination rule in Table2 , the misrouting direction can be either N or S and the fault direction of M is set to E. Suppose that N is selected as the misrouting direction.... In PAGE 14: ... The e-cube hop is blocked again at node (3,3). By the rule in Table2 , the fault direction of M becomes E and the misrouting direction becomes S since that is the only feasible direction. At node (3,3) and (2,3), the... In PAGE 15: ...c1 + 0 - c - 1 c (9,0) (9,9) (0,9) (0,0) (2,0) (7,5) (7,9) (3,8) (6,8) (1,5) Figure 8: Example of fault tolerant routing in 2D mesh misrouting direction and the fault direction of M is determined by the rule in Table2 also. At node (1,3), M becomes normal message and travels up to (1,5) as a normal message.... In PAGE 15: ... The misrouting direction and the fault direction of M is determined by the rule in Table 1 at nodes (1,5) and (2,5). At other nodes, rules in Table2 are used. Consider another message M0 from (7,9) to (3,8).... In PAGE 16: ... Proof. Once a row message is misrouted, the message routes along current fault polygon until it becomes normal again by fault direction determination rule in Table2 . Only after the routing status of the message becomes normal, channels not on current fault polygon can be used.... In PAGE 17: ...column again by rules in Table2 . When the message arrives at the destination column, the message uses its e-cube hop if the hop is not faulty.... ..."

"... In PAGE 4: ... The general process of learning in a neural network, using the standard backpropagation model (Rumelhart et al., 1986), is given by the ve-step algorithm in Table1 . For the details of the backpropagation algorithm, see Rumelhart et al.... ..."