«By Zachary Alexander Rosner A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Psychology ...»
The Generation Effect and Memory
Zachary Alexander Rosner
A dissertation submitted in partial satisfaction of the
requirements for the degree of
Doctor of Philosophy
University of California, Berkeley
Committee in charge:
Professor Arthur P. Shimamura, Chair
Professor William J. Jagust
Professor Matthew P. Walker
The Generation Effect and Memory
Zachary Alexander Rosner Abstract The Generation Effect and Memory by Zachary Alexander Rosner Doctor of Philosophy in Psychology University of California, Berkeley Professor Arthur P. Shimamura, Chair Educators and psychologists have extolled the benefits of active learning techniques such as organizing material, self-explaining, learning through experience, and practicing retrieval for years. Underlying these strategies is the generation effect, an encoding phenomenon in which actively generating rather than passively learning information improves the subsequent retrieval of item information. Despite rather extensive analysis of the generation effect, the processes underlying it are not fully understood. Theories suggest that active generation increases cognitive effort, conceptual processing, item distinctiveness, and semantic processing. Further, generation has also been shown to have varying positive, negative and null effects for contextual features such as order, color, and spatial location, prompting tradeoff and transfer-appropriate processing accounts. This dissertation investigates the positive and negative effects of generation, the universality of the generation effect, and its underlying neural mechanisms. Further, these studies test various explanations of the generation effect, and a transfer-appropriate processing account is considered in detail.
In the first set of studies, I used five experiments to investigate the ways in which active generation can influence memory for item information, related item information, and contextual information. Employing synonym (e.g., ACADEMIC – SCH_L_R), antonym (e.g., question – a____), idiom (e.g., it’s raining cats and ( )), picture, and category-exemplar (e.g., animal – c_t) generation tasks, the positive generation effect for item memory was generally robust, and persisted over long periods of retention and in the face of cognitive distraction. However, negative generation effects were found for font color memory, while null effects were found for background color and location memory. Further, generation was found to impair memory for related items and even the items themselves under certain circumstances.
The second set of studies investigated the degree to which the positive generation effect translates to participants in China, a country that stresses a Confucian rather than Socratic learning style reminiscent of active generation. To address memory for contextual details in a culture that processes information in a field-dependent rather than field-independent manner as in the United States, we also examined the effect of generation on color and spatial location.
American and Chinese participants read or generated idioms (e.g., it’s raining cats and ( ); 倾 家荡( )) presented in different colors or locations, and were tested for item and context 1 memory. For both groups, generation improved item memory. However, American individuals exhibited a negative generation effect only for color memory, while Chinese individuals exhibited negative effects for both color and location memory. These experiments demonstrate the universality of the positive generation effect and the first negative generation effect for location memory to my knowledge.
Finally, I explored the neural basis of the generation effect in an fMRI study. During encoding, participants read or generated synonyms from cues (e.g., GARBAGE – W_ST_). Again, compared to simply reading target words, generating target words significantly improved later recognition memory performance. During encoding, this benefit was associated with a broad neural network that involved both prefrontal (inferior frontal gyrus, middle frontal gyrus) and posterior cortex (inferior temporal gyrus, lateral occipital cortex, parahippocampal gyrus, ventral posterior parietal cortex). These results leave open the possibility that active generation increases attention and cognitive effort (prefrontal and posterior cortical activation), conceptual and semantic processing (IFG and MTG), and item distinctiveness (LOC and ACC).
Overall, active generation proved to be a powerful encoding strategy, engaging a wide range of cognitive processes and broad networks of neural activity. Seemingly an almost effortless task, generation enhanced item memory for various stimuli under several conditions.
However, active generation had limitations, as it impaired both item and context memory in certain situations. I propose that a transfer-appropriate processing account in which active generation promotes conceptual processing and reduces perceptual processing, ultimately enhancing memory for item information, impairing memory for intrinsic contextual information, and ignoring memory for extrinsic contextual information, best accounts for this pattern of positive and negative generation effects.
As I reflect upon my graduate career, I realize that no person is an island. My dissertation is not simply the product of six years of my own work. It is the product of 30 years of support and guidance from a group of people who allowed me to pursue my dreams. Your help has been unconditional, and I am forever in your debt.
First, I would like to thank the National Science Foundation, the National Institute of Neurological Disorders and Stroke, and the National Institute of Health. This research was supported by NINDS Grant P01 NS40813, NSF Grant BCS-0745835, and NIH Grant NS040813.
Additionally, I would like to thank the NSF East Asia and Pacific Summer Institute for allowing me to travel to both China and Taiwan to pursue international collaborations regarding crosscultural research.
Next, I would like to thank my wonderful lab mates and their families, who essentially became my family. This includes Ellen Klostermann, Joel, and Miles Wallace, Jeremy and Erin Elman, Matt Cain, and Diane Marian. From allowing me to sleep on your couches and forcing me to go to the doctor, to the wonderful ideas, collaborations, neuroimaging instruction, and general support and guidance, I cannot imagine having a successful graduate career without you.
I would also like to thank all of my wonderful research assistants and others who have helped me with their efforts. This includes Alexandra Apple, Laurel Brown, Adelle Cerreta, Irene Chen, Eliot Chern, Kara Chung, Brendan Cohn-Sheehy, Ben DeCoudres, Arushi Goonewardene, Amanda Hume, Eba Kim, Sabrina Leu, Disi Li. Victoria Martins, Bita Minaravesh, Fawn Miller, Gloria Park, Miri Faith Park, Alice Nguyen, Ryan Peretz, David Roberts, Rebecca Stevenson, Nancy Tsai, Alice Verstaen, and Elise Vu. I would especially like to thank Stephanie Yang, who helped me to create Mandarin stimuli for, and run, the idiom experiments in China. I have also had amazing mentorship along the way, and I would have never graduated with a Ph.D., let alone made it to graduate school without David Penn, David Roberts, Neil Mulligan, Abigail Panter, Ian Dobbins, Elizabeth Marsh, Nina Gabelko, Marian Diamond, John Kihlstrom, Rich Ivry, Steve Hinshaw, Matthew Walker, William Jagust, Kaiping Peng, and of course Arthur Shimamura.
I would also like to thank my friends Jordan Hutchinson, Will Robinson, Eshanthi Ranasinghe, Amal Saade, Hdimitri Alexander, Ben Couch, Joshua Stalford, Phillip Bush, Rachel Rosenberg, Kevin Uttich, Peter Butcher, Eric Walle, Taraz Lee, Wilbur Putney Williams III, Reno Yeh, Nicholas Song, Barbara Metzenbaum, Shane and Karen Neubert MacCarthy, Kat Fitzpatrick, and Richard Figueroa. I am especially appreciative of my father, who has given me indispensable and practical advice throughout the years, and my brother Tyler, who forces me to behave in a manner consistent with how I tell him to behave, simply so I am not a hypocrite.
Finally, I would like to thank my mother. Thank you for teaching me how to do math in the dirt on the dashboard of our car, for helping me with homework until neither of us understood it, for doing my laundry even in college, for watching television marathons and doing puzzles at 4 in the morning, for following Cleveland sports, for having a great sense of humor and aesthetics, for challenging me when I needed it, for giving me every tool I have ever needed to be successful but more importantly to be good, for giving everything to make sure I that wanted for nothing, for being my inspiration, and for being my best friend.
Without you, I would never have become the person that I am today.
iii Chapter 1: Introduction Introduction Educators have long praised the benefits of active learning techniques such as paraphrasing information, self-explaining, self-testing, and learning through experience. One common underlying aspect of each of these learning strategies is the generation effect, an encoding phenomenon in which actively generated information is retrieved more successfully than passively learned information. Slamecka and Graf (1978) demonstrated the positive effects of active generation on item memory. For example, when generating information during an antonym completion task at encoding, some participants saw the word pair hot-c___, whereas in the read condition others saw hot-cold. In either instance, the word hot was the cue, and cold was the target. Participants demonstrated both better recall and recognition of items that were previously generated rather than read. This positive generation effect for item memory has been demonstrated with verbal information (Slamecka & Graf, 1978), arithmetic problems (McNamara & Healy, 2000; R. W. Smith & Healy, 1998) and pictures (Kinjo & Snodgrass, 2000), and can occur in as little as 250 milliseconds (R. W. Smith & Healy, 1998). Further, actions that are performed rather than observed (Zimmer et al., 2001) and arguments that are self-generated rather than listened to (Petty, 1981) are better remembered. These effects persist for various study paradigms including intentional and incidental learning, and for different testing methods including recognition, cued recall, and free recall (Bertsch, Pesta, Wiscott, & McDaniel, 2007). Additionally, generating information has proven beneficial in older adults (Taconnat et al., 2006; Taconnat & Isingrini, 2004), patients with various traumatic brain injuries, (Lengenfelder, Chiaravalloti, & DeLuca, 2007) and even in patients with mild cognitive impairment or early stages of Alzheimer’s Disease (Souliez, Pasquier, Lebert, Leconte, & Petit, 1996).
Despite rather extensive analysis of the generation effect, the processes underlying it are not fully understood. Theories suggest that active generation increases cognitive effort (McFarland Jr. et al., 1980; Tyler et al., 1979), conceptual processing (Jacoby, 1983), item distinctiveness (Begg, Snider, Foley, & Goddard, 1989; Hunt & McDaniel, 1993; Kinoshita, 1989), and semantic processing (McElroy, 1987; McElroy & Slamecka, 1982), yet none of these views completely explains the phenomenon. Further, generation has been proposed to improve cue-target encoding (Hirshman & Bjork, 1988) and, when using structured lists of related items, inter-target relational encoding (McDaniel, Waddill, & Einstein, 1988). While much evidence exists to support the overarching positive effects of generation, there are limitations. For example, the generation effect is reduced when the read and generate conditions are manipulated between rather than within subjects (Slamecka & Katsaiti, 1987). Additionally, generation appears to have no effect on the recognition of nonwords (McDaniel et al., 1988; Payne, Neely, & Burns, 1986), indicating that the generation effect may only occur for stimuli with preexisting representations (Gardiner & Hampton, 1985; Nairne, Pusen, & Widner, 1985).
Further, there are instances in which generation impairs certain aspects of memory, a phenomenon known as the negative generation effect. For example, the activate generation of items impairs memory for source or contextual features, such as memory for temporal order (Nairne, Riegler, & Serra, 1991), the color and font of presented items (Mulligan, 2004;
Mulligan, Lozito, & Rosner, 2006), and the person who presented items (Jurica & Shimamura, 1999). To explain this negative generation effect, Jurica & Shimamura (1999) proposed an itemcontext trade-off account in which active generation enhances the encoding of item information at the expense of forming contextual associations. Mulligan et al., (2006) on the other hand, adopted a transfer-appropriate processing (TAP) account (see Jacoby, 1983), arguing that active generation promotes relatively more conceptual processing, whereas passive reading allows participants to rely relatively more on perceptual processing. As item retrieval is generally assessed in a conceptual manner, TAP predicts positive generation effects on most standard tests of memory. However, memory for perceptual features, such as the color or font type of presented items, should benefit more from passive study.
This dissertation first describes the history of our understanding of the positive and negative effects of active generation. Then, the following studies aim to add to our knowledge of, and test various theories for, the generation effect. These goals are accomplished through 3 collections of experiments investigating the positive and negative effects of active generation, the universality of the generation effect among different cultures, and the neural mechanisms underlying the generation effect.