Consequences of violating the independence assumption in analysis of variance.

Discusses the biasing effects of nonindependence of observations on the mean squares used to test the effect of some discrete independent variable. Nonindependence of observations is defined, and 3 commonly assumed patterns of nonindependence are identified: nonindependence due to groups, nonindependence due to sequence, and nonindependence due to space. How the bias in both the mean square for treatment and the mean square for error can be derived when each of the 3 patterns of nonindependence is ignored in analyzing the effect of a discrete independent variable is demonstrated. Ways to eliminate the sometimes considerable biases, either by including the source of nonindependence in the analysis, transforming the data to remove it, or modeling it, are discussed. It is concluded that nonindependence of observations should be viewed as a substantive issue central to many areas by psychological research. (28 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Violations of the independence assumption in process dissociation.

L. L. Jacoby, J. P. Toth, and A. P. Yonelinas (1993) advocated a process-dissociation procedure for estimating the contributions to task performance of consciously controlled (R) versus automatic (A) memory processes. The procedure relies on the strong assumption that memory-guided performance attributable to R is stochastically independent of that attributable to A. Violations of this independence assumption can produce artifactual dissociations between estimates of R and A. Such artifactual dissociations were obtained in a series of word-stem completion experiments: R increased with presentation duration, whereas A, paradoxically, decreased. Direct evidence for nonindependence was obtained from correlations between R and A in each of the experiments. These results suggest that the independence assumption was violated, and other applications of process dissociation should not be taken at face value without a thorough evaluation of independence. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Associative priming in word fragment completion: a dissociation between explicit and implicit retrieval processes

Three experiments investigated associative priming in word fragment completion. In associative priming, the study word that acts as a prime is semantically related in some way to the response word that the subject must produce or respond to at test. For example, a prime might be semantically related to the solution to its paired word fragment (e.g. study "VANILLA", solve fragment "-H-C--A-E" at test, solution is "CHOCOLATE"). Associative priming therefore differs from both repetition and conceptual priming, in which the studied primes are themselves the words that must be produced or responded to at test. In Experiment 1, associative primes were found to influence word fragment completion performance on an explicit test, but not on an implicit test. Experiment 2 demonstrated that the effects of associative primes on explicitly instructed fragment completion cannot be attributed to the specific information about cue-prime relationships that is included in the explicit instructions. Experiment 3 demonstrated that a manipulation of modality, a variable known to disrupt implicit retrieval processes, disrupts repetition priming on an explicit test, but not associative priming. The results of these three experiments suggest that whereas repetition primes are retrieved from memory by both explicit and implicit retrieval processes, associative primes are retrieved by only explicit processes. These data suggest that implicit retrieval processes are cue-dependent processes which automatically retrieve memory information that provides a good match to retrieval cues. Explicit retrieval processes are cue-independent, functioning as an intentional retrieval set to access particular categories or types of memory information.     

On the logic of testing the independence assumption in the process-dissociation procedure

The process-dissociation procedure is designed to provide quantitative estimates of the influence of explicit and implicit memory in a variety of tasks. The procedure relies on the assumption that these two forms of memory produce independent influences on performance. Prior investigators have attempted to test this assumption by determining whether the parameter representing the influence of implicit memory (denoted A) is constant across experimental conditions. I argue that the constancy of A cannot provide an appropriate test of the independence assumption, because (1) the prediction of constancy can be generated without the assumption of independence, obviating the need to posit independence; and (2) the constancy of A does not necessarily imply independence, even if one assumes that a dependency hypothesis, supplemented by ancillary assumptions (Curran & Hintzman, 1995), predicts differences in A. I close by emphasizing that we can test the independence assumption by using standard procedures that compare the fit of a model that assumes independence with the fit of a model that assumes dependence.     

Level of processing affects priming in word fragment completion.

In 4 experiments, Ss studied words under learning conditions that promoted semantic or physical processing. An implicit word fragment completion test was administered (e.g., complete l ph t for elephant). When semantic and physical study conditions were manipulated between Ss (Exps 1 and 3) or within Ss in a blocked fashion (Exps 3 and 4), significant levels of processing (LOP) effects were obtained. When semantic and physical conditions were presented in a mixed list (Exps 2, 3, and 4), the LOP effect was smaller and not significant. A survey of the literature on LOP effects in implicit perceptual tests revealed that priming in these tests was consistently greater in the semantic than physical condition, with reports of statistically significant LOP effects. These findings contradict the widely held notion that LOP does not affect priming in implicit perceptual tests and have implications for contemporary accounts of performance in implicit and explicit measures of memory. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Contingent dissociation between recognition and fragment completion: The method of triangulation.

Two experiments conforming to the logic of the method of triangulation were conducted. Following the study of a list of words, the first of two successive tests (recognition) was identical for two groups of subjects, but the second one, in which the same word-fragment cues were presented to both groups, differed with respect to retrieval instructions. Subjects in one group engaged in cued recall of study-list words, whereas those in the second group completed the fragments with the first word that came to mind. Both experiments yielded the same result: The dependency between the first and second tests, indexed by Yule's Q statistic, was greater for recognition and cued recall than it was for recognition and fragment completion. These results speak to the controversial issue of the usefulness of contingency analyses of data from successive memory tests. The results are interpreted in a theoretical framework consisting of an integration of the idea of a hypothetical quasi-memory system with the transfer-appropriate procedural approach. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Consequences and causes of correlations in process dissociation.

In T. Curran and D. L. Hintzman's (see record 1995-42725-001) article, the authors explained how violations of the independence assumption could affect the process-dissociation procedure and presented evidence that was consistent with the hypothesized effects of independence violations. L. L. Jacoby, I. M. Begg, and J. R Toth (see record 84-21424) argued that independence violations could not account for Curran and Hintzman's results. In this reply, the current authors demonstrate that even moderate correlations between recollective and automatic word-stem completion can cause paradoxical dissociations like those the authors previously reported, and they explain how Jacoby et al came to their contrary conclusion. Second, the authors argue that discussion by Jacoby et al of process-dissociation "boundary conditions" is consistent with Curran and Hintzman's original claims. Finally, the authors discuss problems with the evidence for independence that Jacoby et al presented in their Table 1, including their questionable criteria for excluding experiments from the table and the fact that they did not consider statistical power. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Direct comparison of two implicit memory tests: Word fragment and word stem completion.

In 3 experiments, the implicit memory tests of word fragment and word stem completion showed comparable effects over several variables: Study of words produced more priming than did study of pictures; no levels-of-processing effect occurred for words; more priming was obtained from pictures when Ss imaged the pictures' names than when they rated them for pleasantness; and forgetting rates were generally similar for the tests. A different pattern of results for the first 3 variables occurred under explicit test conditions with the same word fragments or word stems as cues. It is concluded that the 2 implicit tests are measuring a similar form of perceptual memory. Furthermore, it is argued that both tests are truly implicit because they meet the D. L. Schacter et al (1989) retrieval intentionality criterion: Levels of processing of words have a powerful effect on explicit versions of the tests but no effect on implicit versions. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Word context during initial exposure influences degree of priming in word fragment completion.

Priming in word fragment completion is revealed by the increased probability of correctly completing a fragment like "_ll_p_e' when the word "ellipse' was seen recently. Three experiments investigated the effects on priming of manipulating the context in which the words were seen. Three principal results emerged. Experiments 1 and 2 demonstrated that there was much more priming for words studied in a to-be-learned list than read in meaningful passages. In these same two experiments, low-frequency words were subject to more priming than were higher frequency words, regardless of context. Experiment 3 revealed more priming for words when they did not fit sensibly into connected discourse than when they did. The results suggest that context plays a critical role in priming: As a word moves from being contextually bound in meaningful discourse to being isolated in a list, its probability of priming increases. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Memory blocks in word fragment completion caused by involuntary retrieval of orthographically related primes.

Seven experiments showed that word fragments are not solved as well following prior exposure to orthographically similar primes (e.g., ANALOGY as a prime for A-L- -GY relative to orthographically dissimilar primes (e.g., UNICORN). This blocking effect was influenced by the modality (auditory vs visual) of the primes but not by the depth to which they were processed. This blocking effect occurred even when participants were informed about it and told to try to avoid remembering the primes, and it was not affected by the proportion of test fragments for which the orthographic primes were correct vs incorrect answers. The results have implications for theories concerned with unconscious mechanisms that underlie memory blocking and blocks to creative problem solving. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

The time course of perceptual and conceptual contributions to word fragment completion priming.

Two experiments examined the time course of the availability of perceptual and conceptual information in priming on the word fragment completion test. Ss encoded primes as either visual words, auditory words, or pictures. In Exp 1, word fragments were exposed for either 500 msec, 1 sec, 5 sec, or 12 sec. Only the visual words produced priming at the 500-msec and 1-sec exposure times. In Exp 2, Ss were allowed up to 20 sec to solve each fragment; response latencies were recorded, and cumulative response curves were generated. Visually primed fragments were solved at a faster rate than either auditorily or pictorially primed fragments. The results suggest that although conceptual processing can contribute to word fragment priming, perceptual processes are recruited earlier and at a faster rate. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

The effect of proportion overlap and repeated testing on primed word fragment completion.

Extended the idea of a proportion overlap (PPO) effect in perceptual identification to a word fragment completion (WFC) test of implicit memory. 80 university students studied a list of words (e.g., cheetah) and received an implicit WFC test (e.g., complete -h--t-h). On the test, the ratio of studied to nonstudied items (PPO) was 0, 25, 50, 75, or 100%. Ss were administered the identical test twice. PPO did not affect priming in WFC, on either the 1st or 2nd test. Also, the completion of studied and nonstudied fragments increased over repeated tests, but priming (the studied–nonstudied rate) remained unchanged. The PPO of items between study and test does not affect performance on primed WFC. (French abstract) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Cue depreciation: When word fragment completion is undermined by prior exposure to lesser fragments.

Studied impairment in ability to think of a previously studied item resulting from a change in extra-item context from study to test in 5 experiments, using a total of 156 Ss (primarily university students). The following results were obtained: a fragment (e.g, r-i--rop) of a just-studied word (raindrop) was shown to be less readily completed if it was presented bit by bit (r------p, r----r-p, r-i--r-p, r-i--rop) rather than all at once (Exps I, III, IV, and V). No such effect was found if the word had not been studied beforehand (Exps II–V). This pattern of results occurred even when fragments of studied and nonstudied words occurred in the same test and under conditions in which Ss could not tell whether a given fragment was of a studied or nonstudied word (Exps IV and V). In addition, for words that had been studied beforehand, the impairment was shown to increase systematically with the number of steps involved in the presentation of the word fragment (Exp III) and also to persist when the time allowed for completion of the final version of the fragment was increased from 4 sec to a full minute (Exp V). The target words are appended. (10 ref) (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Picture fragment completion: Priming in the pigeon.

It has been suggested that the system behind implicit memory in humans is evolutionarily old and that animals should readily show priming. In Experiment 1, a picture fragment completion test was used to test priming in pigeons. After pecking a warning stimulus, pigeons were shown 2 partially obscured pictures from different categories and were always reinforced for choosing a picture from one of the categories. On control trials, the warning stimulus was a picture of some object (not from the S+ or S– category), on study trials the warning stimulus was a picture to be categorized on the next trial, and on test trials the warning stimulus was a randomly chosen picture and the S+ picture was the warning stimulus seen on the previous trial. Categorization was better on study and test trials than on control trials. Experiment 2 ruled out the possibility that the priming effect was caused by the pigeons' responding to familiarity by using warning stimuli from both S+ and S– categories. Experiment 3 investigated the time course of the priming effect. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Levels of representation in visual word recognition: a dissociation between morphological and semantic processing

A robust semantic priming effect typically occurs in visual word recognition if the prime is read before a response to the target. However, this effect is dramatically reduced if a letter search is performed on the prime prior to responding to the target. Three lexical decision experiments document the new observation that morphological priming is preserved following letter search on the prime. This dissociation between morphological and semantic priming following letter search can be understood in the context of an interactive activation framework. In addition, the implications of these results for connectionist and compound cue accounts of word recognition, as well as the issue of automaticity in word recognition, are discussed.     

Priming effects in picture fragment completion: Support for the perceptual closure hypothesis.

The perceptual closure hypothesis says that priming will be optimum when just enough information is available in the prime to support closure. Across 5 experiments, a moderately complete fragmented image (Level 4) produced more priming than an almost complete (Level 7) or a very incomplete (Level 1) fragmented image. Only Level 4 priming was improved by increases in prime duration and by showing the prime again after Ss attempted to identify it. Explicit memory played little role in primed fragment completion except for Level 1 priming, in which specific fragment memory was responsible for the entire effect. In contrast, true perceptual learning was shown to be responsible for Level 4 and Level 7 priming. These priming effects cannot be accounted for by the transfer-appropriate procedures approach of H. L. Roediger and M. S. Weldon (1987) because Level 1 priming produced less transfer to Level 1 identification at test than Level 4 priming did. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Affective valence and memory in depression: Dissociation of recall and fragment completion.

Depressed (n?=?16) and nondepressed (n?=?16) Ss' memory for affectively valenced words was assessed by an explicit test (free recall) and an implicit test (word fragment completion). Under free-recall instructions, depressed Ss recalled significantly more negatively valenced than positively valenced words, whereas the opposite pattern was observed in nondepressed control Ss. These results replicate those previously reported in the literature. The differential effect of word valence was absent, however, when memory was tested implicitly: Depressed and nondepressed Ss exhibited equivalent priming of positive and negative words. These data are discussed in terms of the J. M. Williams et al (1988) model of depression. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

The problem of differing false-alarm rates for the process dissociation procedure: Comment on Verfaellie and Treadwell (1993).

Results reported by M. Verfaellie and J. R. Treadwell (see record 1993-18420-001) contain an interesting paradox: Under standard study conditions in which Ss read words, amnesic patients and control Ss performed identically, both in terms of overall recognition hit rate and when the data were decomposed by L. L. Jacoby's (see record 1992-07943-001) process dissociation procedure into consciously controlled and automatic components of performance. One reason for this curious outcome is that false-alarm rates differed considerably between amnesic patients and control Ss, which is not taken into account in Verfaellie and Treadwell's application of the process dissociation procedure. Considered in this article are 3 possible reactions to the problem of false-alarm rates differing between S groups (or between experimental conditions) for the process dissociation procedure. A correction can be applied either (a) before the process dissociation procedure is used or (b) after the consciously controlled component has been estimated from the procedure. Alternatively, (c) data with this problem may simply be uninterpretable through analysis with the process dissociation procedure. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Visual word recognition: A dissociation of lexical and semantic processing.

These experiments illustrate 2 new dissociations in word-recognition tasks. In one, relatedness facilitated lexical decision but impaired searching for a common letter in the same pairs of words (a cross-over interaction between relatedness and task). In the other dissociation, lexicality facilitated performance (words processed faster than nonwords) while relatedness impaired performance (related words processed slower than unrelated words) in the letter search task. Two classes of explanation are discussed. In the first, the perception of relatedness serves to focus attention to the word level, thereby making explicit letter level processing more difficult and/or increasing the number of competing lexical entries via priming. In the second, spreading inhibition makes related words more difficult to process than unrelated words. (PsycINFO Database Record (c) 2010 APA, all rights reserved)