This study examined the effects of similarity between the processing of acquisition and the processing of test materials on performance in a problem solving task. Previous work by Perfetto, Bransford, and Franks (1983) demonstrated that uninformed subjects' failure to utilize relevant acquisition information in a later problem solving task is the result of a failure to spontaneously access such information. The present study demonstrated that spontaneous access can be enhanced when both acquisition and test materials are processed in a similar manner, that is, in a problem-oriented manner. Furthermore, the present findings indicate that the processing similarity leading to enhanced access is specific to particular acquisition and test items, rather than a general problem solving set induced at acquisition and the subsequent testing situation. Results are interpreted within a transfer-appropriate processing perspective.

Prior research has demonstrated that subjects' failure to spontaneously access acquisition information during a problem-solving task results in performance deficits that persist even after the subjects are informed of the relevance of acquisition information (Perfetto, Bransford, & Franks, 1983). The present experiments examined the hypothesis that these deficits are due to interference from previously generated inadequate solutions to the problems. The results indicated that subjects who generated responses on a first problem-solving trial were able to recall fewer acquisition answers on a second, explicit memory trial than were subjects who read the same Trial! responses as experimenter-provided material. This effect seems to be due, at least in part, to greater spontaneous noticing of the relevance of acquisition information by read than by generate subjects.The manner in which past experiences influence processing of current events is an important concern for the psychology of memory. Current theories of memory have attempted to understand this interaction within what might be called a directed-memory paradigm. A crucial feature of the resulting experiments is that subjects are explicitly directed to complete a memory task with reference to some specified set of previous experiences. They are told what they must remember and where the to-beremembered information was last encountered. Note, however, that such explicit memory cues are seldom present outside of a laboratory setting.Recently, a number of studies have used an alternative spontaneous-access paradigm (e.g., Gick & Holyoak, 1980;Perfetto, Bransford, & Franks, 1983;Weisberg, DiCamillo, & Phillips, 1978). In this paradigm, subjects are presented with problems that can be solved by using recently experienced information. However, they are not explicitly directed to use this information. These studies suggest that spontaneous access of information is not an automatic process, even when the information is both highly relevant and potentially available .. Furthermore, these studies indicate that conclusions drawn from experiments conducted under explicit retrieval conditions are not necessarily generalizable to situations that require spontaneous memory access.In previous work (perfetto et al., 1983), we tested subjects' spontaneous memory for acquisition information by presenting them with word problems that could be unambiguously solved by using statements presented during an incidental acquisition task. Half of the subjects (inThis research was supported in part by Contract MDA-903-84-C-021S from the US Army Research Institute to Jeffery Franks and John Bransford. We would like to thank Jane Kasserman. Larry Jacoby. Fergus Craik, and two anonymous reviewers for comments on earlier versions of this manuscript. Requests for reprints should be sent to Greg Perfetto, Department of Psychology. Vanderbilt University, 134 Wesley Hall, Nashville, TN 37240. formed) were told that the acquisition items would be helpful; the remaining subjects (uninformed) were n...

The proportion of contextually related priming pairs was manipulated in a perceptual identification task to determine the relative contributions of automatic and controlled processes to performance. Subjects studied a set of sentences and then identified prime-target pairs consisting of (1) two words from the same studied sentence (related context), (2) two words from different studied sentences (unrelated context), or (3) a nonword prime and a word from a studied sentence (neutral context). When the identification test list included a high proportion of contextually related pairs, facilitation for related-context pairs and inhibition for unrelated-context pairs were observed relative to comparable nonstudied pairs. In the low-proportion condition, performance was unaffected by the proportion manipulation. Based on these results, we argue that contextual effects in perceptual identification reflect the operation of controlled processes. There is substantial evidence that prior experimental facilitate identification. The procedures for this experipresentation of words can promote better perceptual iden-ment were adapted from Ratcliff and McKoon (1978). In tificationof those words in a subsequenttest (e.g., Jacoby, their work, subjects were presented with a set of sentences 1983; Jacoby & Dallas, 1981; Jacoby & Witherspoon, and were subsequently given a recognition reaction-time 1982). Jacoby and colleagues argued that this facilitation test for a list of words, some of which were old words in performance is largely due to enhanced perceptual that had appeared in the sentences, whereas others were fluency for the previously experienced items, with per-new words. Reaction times were faster for old words that ceptual fluency being essentially related to the sensory followed other old words from the same clause or senproperties of the words. Sensory-based perceptual fluency tence. On the basis of subsequent experiments , Ratcliff appears to be the major process underlying perceptual and McKoon (1981) concluded that these priming effects identification performance when the situation involves are due to automatic processes. relatively unorganized lists of independent words during A method that is typically used to differentially manipuinitial exposure and later test. However, Jacoby also late automatic versus controlled processes involves varyspeculated that semantic processes can contribute to per-ing the proportion of target items that instantiate the receptual identification performance in cases in which con-lation that is of central interest. For example, Ratcliff and textual factors are available to support such meaning-McKoon (1981, Experiment 1) varied the proportion of related processes. For example, Jacoby (1983, Experi-appropriate priming pairs in the recognition test. The ment 1) demonstrated that perceptual identification varies degree of priming observed should be affected by this as a function of the proportion of old words in the test proportion manipulation if priming is at least partially a list. In add...