"... In PAGE 4: ...ignificant conditioning (mean conditioning score = 6.6 * 4.5). We have replicated this basic effect in three additional experi- ments with the same design but with different intertrial intervals or numbers of training trials (see below and Table3 ). In each case the result was the same: Animals receiving normal training showed significant conditioning, while animals receiving addi- tional, unpredicted USs did not.... In PAGE 5: ... In order to do this, we had to alter two other variables: the number of USs and the time between USs. To test whether these alterations, rather than the alteration in contingency, might account for the observed decrease in conditioning, we performed three additional exper- iments ( Table3 ). First, we were concerned that animals given unpredicted USs might not have conditioned because they re- ceived a higher frequency of USs (one every 2.... In PAGE 5: ... However, US frequency does not appear to be a critical variable, since (1) animals given normal training with a 2.5 min IT1 showed good conditioning ( Table3 , Experiment B), and (2) animals given unpredicted USs with a 10 min ITI-thus receiving a US once every 5 min, on average-still did not condition (Table 3, Ex- periment D). Second, it was possible that animals given unpre- dicted USs did not condition because they received more USs (10) than animals given normal training (five).... In PAGE 5: ... However, US frequency does not appear to be a critical variable, since (1) animals given normal training with a 2.5 min IT1 showed good conditioning (Table 3, Experiment B), and (2) animals given unpredicted USs with a 10 min ITI-thus receiving a US once every 5 min, on average-still did not condition ( Table3 , Ex- periment D). Second, it was possible that animals given unpre- dicted USs did not condition because they received more USs (10) than animals given normal training (five).... In PAGE 5: ... On average, animals receiving normal training showed significant differential conditioning, while animals receiving unpre amp;tedUSs showed no differential conditioning. mals given 10 trials of normal training-and thus 10 USs-also showed good conditioning ( Table3 , Experiments C and D), which indicates that an increased number of USs per se does not interfere with conditioning. These results suggest that the decrease in conditioning by animals receiving unpredicted USs in each of these experiments was not due entirely to differences in US frequency or number, but rather, at least in part, to the difference in the degree to which the US was contingent on the CS+.... ..."

"... In PAGE 48: ... Over 40 % of decision-makers and more than half of those at the operational level (53.3 %) also felt insufficiently informed (see Table1 0). ... In PAGE 54: ...2 %). In other words, in over half of the organisations in question, quality measures are perceived by respondents only at a very abstract level (an organisational goal) or as an implicit requirement ( Table1 1). Only 13 % of respondents stated that methods and instruments were used which were explicitly aimed at ensuring quality in e-learning products and processes.... In PAGE 68: ... Among e-learning providers, quality approaches were used in almost 77 % of organisations, while the proportion was less favourable among e-learning users, although still relatively high (a good two thirds of those respondents). A detailed analysis showed that in over half of the organisations in question, quality measures were perceived by respondents only at a very abstract level (an organisational goal) or as an implicit requirement ( Table1 0). Only 13 % of respondents stated that methods and instruments were used which were explicitly aimed at ensuring quality in e-learning products and processes.... ..."

"... In PAGE 6: ... An Environment with a Large Low Obstacle. In this experimental scenario, we put a single large low obstacle with dimension (100; 100; 30) in the empty environment to evaluate the obstacle avoidance behavior of the formations when the obstacle can be jumped (see Table2 ; simulation steps range from 674 for the Line to 775 for the Snake). The Line behaves good but it still shows some latency in the followers.... ..."

"... In PAGE 19: ... 11. Table3 is a summary of the results of applying the obstacle detection algorithm to the data sets collected at different times of day. The variability of detection is a result of the following factors: (1) Data acquisition rate (1 image per second).... ..."

"... In PAGE 9: ... With the circle model (a disk with radius 400 cm), we expect a point within the cir- cle to be associated with event detection and a point beyond the circle range not to be associated with event detection. After repeating the training test many times, we obtained ir- regularity and training confidence as shown in Table1 . They were calculated for all points associated with training events as follows: Irregularity = n1 +n2 n3 where n1 is number of points inside the circle the events of which are not detected, n2 is number of points outside the circle the events of which are detected, n3 is number of points inside the circle.... ..."

"... In PAGE 21: ... They are modernizing their facilities and are also participating with foreign pharmaceutical industries in programs of technological transfer for the production of newly introduced vaccines, such as HIB and influenza (Brazil 2002). Epidemiological and policy estimates for potential target groups The estimates by this study of the size of potential target groups for HIV/AIDS vaccination in Brazil by relative riskiness of behavior and accessibility ( Table9 ), are based on several large studies supported by the National AIDS Program. They can be found at the site www.... In PAGE 23: ... 19 The potential target groups included in Table9 can be grouped in three categories: those at high risk, due to relative riskiness of behavior, and low accessibility (1-7); those with high accessibility, due to institutional access, and at medium to lower risk (8-13); and the sexually active population (15-49), with a potential although lower risk for HIV/AIDS, and medium accessibility. These groups are plotted according to their ease of access and the number of infections averted by an AIDS vaccine in Figure 1.... ..."

"... In PAGE 24: ... In contrast, firms paying dividends are more likely to be large than firms not paying dividends. The results of the three sample splits for our benchmark ADL model are shown in Table9 ; Table 10 also reports results for the ECM model. For each regression, the second column reports the coefficient on each variable interacted with a dummy assuming value 1 for that particular sub-group of firms, representing the differential coefficients for the same sub-group.... ..."

"... In PAGE 24: ... In contrast, firms paying dividends are more likely to be large than firms not paying dividends. The results of the three sample splits for our benchmark ADL model are shown in Table 9; Table10 also reports results for the ECM model. For each regression, the second column reports the coefficient on each variable interacted with a dummy assuming value 1 for that particular sub-group of firms, representing the differential coefficients for the same sub-group.... ..."

"... In PAGE 7: ... It is instructive to calculate values for a typical highway scenario. Table1 : Typical Values for Highway Obstacle Detection Variable Meaning Value Given (G) / Calculated (C) v vehicle velocity 26.7 m/s (60 mph) G adec max.... In PAGE 14: ... Although some of our current laser sen- sors have a larger instantaneous field-of-view (IFOV) than required by our acuity constraint (see Table 2 on page 12) of 0.1 degrees (as listed in Table1 on page 4) to detect a 20 centimeter obstacle at 65 meters, we believe that we can scale down our experiments appropriately with these sensors. Even without the necessary acuity, a laser with a large spot size may be able to detect an obstacle since a larger spot averages the reflectance signal over a larger area.... ..."