0 1998 OPA (Overseilr Publ~rhcrr Asrocintion) N.V. Rcprintr uvilihble directly from ths publisher Pvblirhcd by lien* under Photocopying pcrmitlcd by liccnsc only the Gordon und Breach S c m a Publirhcrr impnnt. The automated highway system (AHS) is "the application of vehicle/highway automntion technology to provide signficunr improvements in traflic safety, throughput, travel times, comfort, convenience, energy, and [the] environmenl." Under a contract with the Federal Highway Administration to investigate human factors aspects of the design of the AHS, 14 experiments and several analytical tasks have been conducted. The results of the experiments and analytical work on automated-steering failures are reviewed, followed by a summary of the implications of that work for automated highway systems. Then, several critical human Pdctors research issues -from a list of 62 developed during the contract -are presented and discussed.

The present experiment tested several predictions from the Rescorla-Wagner (1972) model of Pavlovian conditioning. For two groups, each of two CSs was independently paired with shock until asymptotic suppression developed. They were then presented simultaneously for a second phase: for one group, US intensity was not changed between phases, while for the second group it was increased. The compound lost associative strength in the first group but not in the second. For a third group, the first phase was run using only one CS. During compound conditioning, the same US was used to reinforce the compound formed by adding a novel CS; no change in suppression was seen during compound conditioning. Other predictions were made concerning additive summation, and these were confirmed, as were predictions concerning a test phase in which the CSs were again presented singly. The evidence is taken as strong support for the Rescorla-Wagner model.According to a recent model of Pavlovian conditioning proposed by Rescorla and Wagner (1972), the change in associative strength (A V) accruing to a conditioned stimulus (CS) as the result of pairing with the unconditioned stimulus or us (Le., a reinforced trial), is determined by the discrepancy between the asymptotic associative strength the US will support and the current associative strength of the stimulus complex in which the CS is embedded (Vtotal) ' The discrepancy between the asymptote and Vtotal determines the direction and magnitude of AV: (a) if this discrepancy is positive, CS will increase in associative strength on that trial; (b) if this discrepancy is negative, CS will decrease in associative strength; and (c) if this discrepancy is zero, no change in associative strength will result.Given these basic rules, it is possible to imagine To answer this question, we first need some rule for determining the associative strength of a compound, given the strengths of its elements. The simplest assumption is that the associative strengths add algebraically, so that VA + VB = V AB; this is the assumption made by Rescorla and Wagner (1972). Given this, two predictions then follow: (a) the total associative strength of the AB compound should exceed that of the individual elements, and AB should initially have greater associative strength than either element alone; and (b) if we continue to reinforce AB with the same US, one or both elements should lose associative strength. This second follows because, should V AB exceed the asymptote which the US will support, the discrepancy between asymptotic associative strength and V AB would be negative, and further reinforcements should lead to decrements in the associative strength of the compound. The model does not demand that both CSs lose associative strength to the same degree, because the magnitude of A V on any trial is also assumed to be a function of stimulus salience. Thus, depending on the relative saliences of A and B, one or both should come to control less suppression if this procedure is carried out long enough. This predi...

No abstract

No abstract

No abstract