«By Zachary Alexander Rosner A dissertation submitted in partial satisfaction of the requirements for the degree of Doctor of Philosophy in Psychology ...»
For the recognition test, participants viewed a series of 80 randomly ordered words (40 target words, 40 new words) in black font. Item and source recognition memory were tested in a 3-alternative, multiple-choice test in which participants decided for each test word whether it was old and previously presented in red (R), old and previously presented in green (G), or new (N).
32 The recognition test was self-paced with the next test trial initiated after each recognition response.
Results Item recognition performance (see Figure 2.1C) was based on responses of “old” to target words, regardless of source (i.e., color memory) accuracy. We subjected item recognition performance to a 2 x 2 ANOVA with encoding condition (generate vs. read) and cultural groups as variables. There was a main effect of encoding type, with the generation condition producing significantly better item recognition than the read condition (F(1,95) = 36.43, p.001). While there was a main effect for culture (F(1,95) = 6.11, p =.02), the positive generation effect was comparable across cultures as the culture x encoding condition interaction was not significant (F(1,95) =.01, p =.91). The positive generation effect was significant in each culture.
Specifically, American participants correctly recognized 69% of generated items and 57% of read items, t(49) = 4.17, p.001, and Chinese participants correctly recognized 61% of generated items and 49% of read items, t(47) = 4.38, p.001, which amounted to a boost of 12% in item recognition for both groups as a result of generating target words. False alarm rates (i.e., identifying a new item as “old”) were comparable across groups (10% for USA participants, 12% for Chinese participants).
As shown in Figure 2.1C, participants exhibited negative generation effects for source (color) memory (F(1,95) = 12.40, p =.001). As with item recognition, there was a main effect for cultural group (F(1,95) = 4.54, p =.04), but no encoding condition x cultural group interaction (F(1,95) =.02, p =.89). Simple effects showed significant negative generation effects for both cultures. Among American participants, color accuracy was 59% in the generate condition and 68% in the read condition, t(49) = 2.65, p =.01, while for Chinese participants, color accuracy was 52% in the generate condition and 61% in the read condition, t(47) = 2.35, p =.02, a 9% drop in source memory for generated words compared to read words in both cultural groups.
Experiment 2.2 (Idiom Location) Cross-cultural Differences in Cognition Findings from the first experiment demonstrated comparable generation effects between American and Chinese individuals. Specifically, both groups exhibited positive generation effects for item memory and negative generation effects for color memory. Yet, as mentioned earlier, memory for some contextual features such as the spatial location of study items often fail to show negative generation effects (Marsh, 2006; Mulligan, 2004). Mulligan (2011) argued that generation impairs memory for intrinsic contextual features, while leaving extrinsic contextual features unaffected. It is possible, however, that cultural differences in cognition, such as fielddependence, lead to different types of contextual features being processed either intrinsically or extrinsically.
According to Markus and Kitayama (1991), the East-Asian definition of the self is construed as interdependent with others and leads to a holistic cognitive style in which one constantly scans the environment for information (field dependence). As described by Nisbett and colleagues, East Asians are “…holistic, attending to the entire field and assigning causality to it, making relatively little use of categories and formal logic” (Nisbett, Peng, Choi, & Norenzayan, 2001, page 291). In contrast, the American definition of the self is construed as independent of others, resulting in a more analytic cognitive style in which one focuses on target 33 objects at the expense of scanning the environment (field independence). Nisbett and colleagues describe this Western analytic nature as “…paying attention primarily to the object and the categories to which it belongs (Nisbett, Peng, Choi, & Norenzayan, 2001, page 291). For example, when asked to photograph a portrait of another person, American participants took close-up shots, capturing only the individual, whereas Japanese participants took more inclusive photographs, allowing for greater presence of background and context (Nisbett & Masuda, 2003).
Numerous studies have illustrated these differences in field dependence. In one experiment, participants viewed a box with a short line drawn from the top, and were asked to redraw the line to either its absolute or relative length in a different sized box (Kitayama, Duffy, Kawamura, & Larsen, 2003). While American participants made smaller errors in the absolute condition, Chinese participants made smaller errors in the relative condition. In a later neuroimaging study, Hedden, Ketay, Aron, Markus, and Gabrieli (2008) revealed that participants had greater parietal and frontal lobe activation when performing their culturally more difficult task, indicating the need for more cognitive control. Further, attention and memory differences occur when viewing focal objects surrounded by complex scenes. Masuda and Nisbett (2006) found that Japanese participants noticed fewer focal and more background changes in a change blindness study. Masuda and Nisbett (2001) also found that Japanese participants were faster and more accurate when recognizing focal objects paired with previously presented backgrounds as compared with novel backgrounds, whereas American participants were unaffected by background status. Extending these findings, eye-tracking revealed that not only did American participants look at the focal object more than 100 milliseconds sooner than did Asian participants, after 500 milliseconds, American participants made longer fixations on the focal object, while Asian participants made more saccadic eye movements to the background (Chua, Boland, & Nisbett, 2005). Even at the neural level, Gutchess, Welsh, Boduroĝlu, and Park (2006) found that Americans had greater activity in object processing regions of the brain, including bilateral middle temporal gyrus, left angular gyrus, and right supramarginal gyrus. As far as location is concerned, it is argued that Chinese people truly internalize location, as in the past people often identified the self as where they came from (Hsu, 1981).
Given these cross-cultural differences in the way in which East-Asians and Americans process focal objects in relationship to the environment, cultural differences for the way in which generation influences location memory may exist. In the first set of experiments, there was a negative generation effect for memory for color. Mulligan (2011) argued for an intrinsic/extrinsic TAP account of the generation effect in which generation promotes conceptual processing, benefiting later item recognition. Passive study, on the other hand, promotes perceptual processing, which benefits memory for intrinsic contextual details such as font color and leaves extrinsic contextual details such as location unaffected. While location is likely processed as an extrinsic contextual detail among American participants, it may be processed intrinsically among Chinese participants. To the extent that Chinese individuals process location as an intrinsic contextual detail, an intrinsic/extrinsic TAP account (Mulligan, 2011) predicts a negative generation effect for location similar to that of color. Experiment 2.2 therefore sought to test whether a negative generation effect exists for location memory among Chinese participants in the absence of such an effect among American participants.
Materials and Methods Participants 34 Forty-eight UC Berkeley undergraduate students participated for 1 hour of research participation credit for partial fulfillment of a psychology course requirement. Thirty-four undergraduate students from Beijing University and Tsinghua University in Beijing, China participated in exchange for ¥15 RMB (approximately 2 US dollars).
Materials All materials were identical to Experiment 2.1, except that location, rather than color, was manipulated for the source memory test. Half of the idioms were presented on the top of the screen, and the other half were presented on the bottom of the screen (Figures 2.2A, 2.2B). Both encoding strategy (generate vs. read) and location (top vs. bottom) were manipulated within participants and counterbalanced so that each idiom appeared in each possible combination of conditions with equal frequency.
Procedure The procedure was identical to that of Experiment 2.1, except that participants were instructed that some of the idioms would be presented at the top of the computer screen, and the others at the bottom of the computer screen. Participants were instructed to remember both the idiom and its location for a later memory test. Item and source memory were assessed using a 3alternative, multiple-choice test in which participants were presented test words in the middle of the screen and decided whether each word was old and had been previously presented on the top half of the screen (T), old and previously presented on the bottom half (B), or was a new word (N).
Results Figure 2.2C displays generation effects for item and source (spatial location) recognition.
As in Experiment 2.1, there was an overall positive generation effect for item recognition (F(1,80) = 44.24, p.001). American participants correctly recognized 71% of generated items and 60% of read items, t(47) = 5.36, p.001. Chinese participants correctly recognized 58% of generated items and 46% of read items, t(33) = 4.12, p.001. Both groups exhibited comparable boosts in performance as a result of generating target words (USA participants = 11% boost, Chinese participants = 12% boost). False alarm rates (i.e., identifying a new item as “old”) were comparable across groups (6% for USA participants, 9% for Chinese participants). Again, the main effect of group (F(1,80) = 14.80, p.001) was significant, while the condition type x group interaction (F(1,80) =.09, p =.76) was not.
The pattern of performance for source memory of spatial location was entirely different from the findings of color memory observed in Experiment 2.1. For American participants, there was a nonsignificant positive generation effect, as source accuracy was 73% in the generate condition and 68% in the read condition t(47) = 1.80, p =.08. However, Chinese participants exhibited a significant negative generation effect, with source accuracies of 66% in the generate condition and 75% in the read condition t(33) = 2.56, p =.02. This pattern was confirmed by a significant encoding condition x cultural group interaction (F(1,80) = 10.34, p.01) and nonsignificant main effects for encoding condition (F(1,80) = 1.14, p =.29) and group (F(1,80).01, p =.98).
35 In general, American participants exhibited better overall item memory than Chinese participants. As it was not entirely possible to control for the exact target words used across cultures, such effects may be attributable to item effects. More importantly, there was a strong, consistent positive generation effect for item memory for both American and East-Asian participants, demonstrating the universal benefit of active generation as a powerful and robust mnemonic technique. Thus, despite potential differences in learning styles, active generation enhanced item recognition performance.
Interestingly, there were some differences between groups in the manner in which contextual information was remembered. Across cultures, a negative generation effect was observed for color memory. Yet, memory for the spatial location of items differed between cultures as a function of active generation. Among American participants, generation had no significant effect on location memory, a finding that has been previously observed (Mulligan, 2004; Mulligan et al., 2006). Among Chinese participants, however, generation significantly impaired location memory to a similar degree as color memory. To our knowledge, a negative generation effect for spatial context has never been reported. This finding, however, helps to explain the inconsistency of the generation effect that has been observed for location memory.
As stated by Mulligan (2004), a TAP account predicts that active generation promotes conceptual processing, whereas reading promotes perceptual processing. In turn, generation benefits performance on conceptual retrieval tasks such as item recognition. However, generation impairs performance on perceptual tasks such as the retrieval of intrinsic contextual information. Memory for extrinsic contextual information, on the other hand, is unaffected (Mulligan, 2011). Among both cultures, color is likely an intrinsic contextual detail, explaining why color memory may benefit from perceptual processing during encoding. This story is entirely different for location, however. American participants are field-independent, ignoring the environment when analyzing a focal object. For this reason, American participants likely process location as an extrinsic contextual detail. However, Chinese participants are more fielddependent (Markus & Kitayama, 1991) and may process location as an intrinsic contextual detail, similar to the way in which both groups process color information, resulting in a negative generation effect.
Finally, this study demonstrated that while the similarities in cognition between cultures greatly outweigh the differences, differences do exist. Even basic cognitive functions such as perceptual memory appear to be influenced by culture. Interestingly, these differences likely occur during encoding. The generation effects were calculated by comparing generate versus read conditions within each culture. Therefore, unless one argues that the qualitative manner in which information is retrieved was affected by the way in which it was initially encoded, any observed effects cannot be the product of cultural differences in representational or retrieval biases, as these were held constant.
Figure 2.1 – (A) Encoding task for Experiment 2.
1 (USA). (B) Encoding task for Experiment 2.1 (China). (C) Behavioral results for Experiment 2.1 (USA and China).
37 Figure 2.2 – (A) Encoding task for Experiment 2.2 (USA). (B) Encoding task for Experiment 2.2 (China). (C) Behavioral results for Experiment 2.2 (USA and China).