«By Zachary Alexander Rosner A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Psychology ...»
13 Experiments 1.4A and 1.4B (Picture Fragment Completion: Picture-Picture; Picture Fragment Completion: Word-Picture) While negative generation effects for context memory such as color have been documented, negative generation effects for item memory are less common. These experiments were motivated by the TAP account’s predictions of positive and negative generation effects as influenced by relative amounts of conceptual and perceptual processing. Using line drawings of common objects, complete (read condition) and fragmented (generate condition) pictures and words were presented at encoding, while complete pictures were presented at test. Previously, Kinjo and Snodgrass (2000) presented complete and fragmented pictures during study and found that both recall and source memory (defined as memory for whether the picture was initially presented in complete or fragmented form) was better if the picture was initially presented in fragmented form. When presented with complete or fragmented pictures during a recognition task, however, performance was best when viewing complete pictures at study and test rather than when viewing fragmented pictures at study and test. This result may be due to similar perceptual processing during both study and test. Further, when presented with a more conceptual retrieval task, such as the name of the picture rather than the picture itself, there was a positive generation effect.
However, if a person is tested with the same form of a picture as during study, then utilizing perceptual processing may be more beneficial. In these two experiments, participants viewed either pictures or words at study and then were tested with the picture. A TAP account would predict that if presented with a picture at both study and test, there should be a negative generation effect for both item and color memory. Active generation should promote more conceptual processing, but in this instance the test of both item and source information is perceptual in nature. When presented with a word at study and a picture at test, however, participants should rely more on the concept of what was seen rather than the consistent processing of perceptual information, which should result in a positive generation effect for item memory. Color memory, however, is still a perceptual task, and should benefit from the read condition at study.
Materials and Methods Participants Ninety-seven UC Berkeley undergraduate students participated for 1 hour of research participation credit for partial fulfillment of a psychology course requirement. Forty-eight participated in the Picture-Picture experiment and 49 participated in the Word-Picture experiment.
Design and Materials Encoding stimuli consisted of 40 pictures derived from the Snodgrass and Vanderwart standardized picture set (Snodgrass & Vanderwart, 1980) of simple line drawings or the corresponding words used to describe them. For the Picture-Picture experiment, half of the pictures were presented in complete form, and the other half were presented in fragmented form.
For the Word-Picture experiment, half of the words were presented in complete form (e.g., FLOWER), while the other half were presented in fragmented form, meaning they were presented with their vowels removed (e.g., FL_W_R). Additionally, for each experiment half of 14 the stimuli were presented in red, while the other half were presented in blue. Both encoding strategy (generate vs. read) and color (red vs. blue) were manipulated within participants and counterbalanced such that each picture appeared in each possible combination of conditions with equal frequency. Only pictures and words that were identifiable with at least 95% accuracy (demonstrated through prior experiments to pilot stimuli) were used. The distractor task consisted of a worksheet of 162 simple arithmetic problems, including addition, subtraction, multiplication, and division. The recognition portion of the experiment contained 80 randomly ordered pictures. Forty of these items were old, consisting of items previously presented during the encoding phase, while 40 were new, consisting of unused items. Stimuli appeared as old and new with equal frequency over all participants.
Procedure Participants were either told they would see a series of pictures (Picture-Picture experiment) or words (Word-Picture experiment). Either way, all participants were told that some of the items would be complete, and others would be fragmented. Additionally, they were instructed that some of the items would be presented in blue, and the others would be presented in red. Regardless of condition, they were asked say the name of each item aloud. This ensured that the correct item was generated, enabling the elimination of incorrectly identified items from future analyses. Participants were also told to remember both the item, and its color, for a later memory test. Before beginning the encoding phase, 2 practice encoding trials (1 complete and 1 fragment) were performed to ensure that participants sufficiently understood the task.
Participants then viewed a series of 40 randomly ordered items. Each trial began with a 2-second fixation inter-trial interval, followed by the presentation of the item for 3 seconds (Figure 1.3AC). Following the encoding portion of the experiment, participants performed the math distractor task. They were asked to answer as many of the problems as they could in 2 minutes. The purpose of the distractor task was to prevent the rehearsal of recently presented pictures, and ensure that long-term memory would be tested.
During retrieval, participants viewed a series of 80 randomly ordered pictures. Forty of these items were old, consisting of items that were previously presented during the encoding task, and 40 were new, consisting of unused items. Participants decided if each item was old or new with a confidence rating (1 = definitely old, 2 = probably old, 3 = probably new, 4 = definitely new). If the item was determined to be old, the participant then decided in which color the item was previously presented with a confidence rating (1 = definitely blue, 2 = probably blue, 3 = probably red, 4 = definitely red). Words were presented one at a time, in black, in the center of the screen.
Results and Discussion In the Picture-Picture experiment, participants correctly recognized 86% of generated items and 93% of read items, t(47) = 3.60, p.001. Color accuracy was 74% in the generate condition and 81% in the read condition, t(47) = 2.05, p.05 (Figure 1.3D; Table 1.1). In the Word-Picture experiment, participants correctly recognized 90% of generated items and 74% of read items, t(48) = 9.10, p.001. Color accuracy was 58% in the generate condition and 65% in the read condition, t(48) = 2.79, p.01.
For item recognition, when a person was presented with a picture at both study and test, there was a negative generation effect for both item and color memory. While the generation task still promoted conceptual processing, the retrieval task was more perceptual in nature, and seeing 15 rather than generating the picture at study resulted in better item and color memory. When presented with a word at study and a picture at test, however, participants could no longer rely on similar perceptual processing during retrieval, thus driving the positive generation effect for item memory. Color memory, however, still relied on perceptual processing, as it is unlikely that participants conceptually processed color information during encoding, resulting in a negative generation effect. While these results do not rule out an item-context tradeoff account (Jurica and Shimamura, 1999), they are more consistent with a TAP account (Jacoby, 1983; Mulligan et al., 2006). These experiments demonstrated that the negative generation effect is not a necessary consequence of a positive generation effect for item memory. Specifically, when encoding and retrieving picture stimuli, generation impaired both item and color memory.
Experiment 1.5 (Category Retrieval Blocking) The previous experiments demonstrated that active generation can negatively impact memory both for the item and its context. Can such negative influences extend to memory for other items? Hirshman and Bjork (1988) found that when using structured lists of related items, generation benefited memory for other target items through inter-target relational processing.
However, it is possible that that the enhanced memory of generated items may impair memory for related items to the extent that generation fails to activate these items. Previously, Anderson, Bjork, and Bjork (2000) found that retrieving category exemplars strengthens memory for those items while weakening memory for unretrieved exemplars, a phenomenon known as retrievalinduced forgetting (Anderson, Bjork, & Bjork, 1994). Indeed, Bäuml (2002) found that the act of self-generating exemplars may impair the recall of previously presented exemplars. However, questions remain. Does the positive effect for generated items interfere with the non-generated related items? If generation does impair memory, does this impairment operate proactively as well as retroactively? Therefore, the purpose of this experiment was to investigate potential negative generation effects on other previously learned information through a task similar to retrieval-induced forgetting (Anderson et al., 1994). During an initial encoding task, participants read or generated category-exemplar word pairs (e.g., FURNITURE – CH_ _R or ANIMAL – DOG). Then, during a second encoding task, participants saw the same categories paired with new read (e.g., FURNITURE - TABLE) or generate (e.g., ANIMAL – C_T) exemplars. Generation enhances item memory to such a great extent that it is possible that this memory strength can block the retrieval of previously learned items through a mechanism similar to that of retrievalinduced forgetting.
Materials and Methods Participants Sixty UC Berkeley undergraduate students participated for 1 hour of research participation credit for partial fulfillment of a psychology course requirement.
Design and Materials Encoding phase 1 stimuli consisted of 60 category-exemplar word pairs (e.g., FURNITURE – CHAIR). Encoding phase 2 stimuli consisted of the same 60 categories from encoding phase 1 associated with new exemplars (e.g., FURNITURE – TABLE). Stimuli were either presented as complete in the read condition (e.g., ANIMAL - CAT) or fragmented, with the vowels removed, in the generate condition (e.g., ANIMAL – C_T). Twenty of the items were presented in the read condition in both study phases, twenty were presented in the read condition 16 during encoding phase 1, and the generate condition during encoding phase 2, and twenty were presented in the generate condition in encoding phase one, and the read condition in encoding phase 2. The encoding strategy (Read-Read, Read-Generate, Generate-Read) was manipulated within participants and counterbalanced such that each category-exemplar pair appeared in each condition with equal frequency. Only well-known category-exemplar pairs in which participants were able to correctly generate the exemplar with at least 99 percent accuracy (demonstrated through prior experiments to pilot stimuli) were used. The distractor task consisted of a worksheet of 162 simple arithmetic problems, including addition, subtraction, multiplication, and division. The retrieval task consisted of the 80 previously presented categories. Directly beneath each category, participants were asked to type both related exemplars.
Procedure Participants were seated facing a computer and told that they would see a series of category-exemplar word pairs. Some of these word pairs would be complete, while for others the exemplar would have its vowels removed. Regardless of condition, they were asked to make a keypress of “1” if they could identify the exemplar. This ensured that the correct exemplar was generated, enabling the elimination of incorrectly identified items from future analyses.
Participants were told to remember the category-exemplar word pair for a later memory test.
Before beginning the encoding phase, 2 practice encoding trials were performed to ensure that participants sufficiently understood the task. Participants then viewed a series of 60 categoryexemplar word pairs. Each trial began with a 2-second fixation inter-trial interval followed by the presentation of a category-exemplar pair for 3 seconds (Figure 1.4A, 1.4B). Following the first encoding portion of the experiment, participants performed a math distractor task. They were asked to answer as many of the problems as they could in 2 minutes. The purpose of the distractor task was to prevent the rehearsal of recently presented word pairs. Next, participants viewed a series of the same 60 categories from the first encoding phase paired with new read or generate exemplars. Following the second encoding portion of the experiment, participants performed another math distractor task. During retrieval, participants viewed the series 60 categories in random order. Participants were asked to respond by typing both previously associated exemplars.