Sound Research WIKINDX
Sound Research WIKINDX |
|
 |
|
Beaty, R. E., & Silvia, P. J. (2012). Why do ideas get more creative across time? An executive interpretation of the serial order effect in divergent thinking tasks. Psychology of Aesthetics, Creativity, and the Arts, 6(4), 309–319. Added by: alexb44 (5/12/26, 4:46 AM) Last edited by: alexb44 (5/12/26, 5:09 AM) |
 |
| Resource type: Journal Article Peer reviewed Published DOI: 10.1037/a0029171 BibTeX citation key: Beaty2012 Email resource to friend View all bibliographic details  |
Categories: General Keywords: Creativity, Divergent thinking, Ideation, Serial order effect Creators: Beaty, Silvia Collection: Psychology of Aesthetics, Creativity, and the Arts |
Views: 3/71
|
| Abstract |
|
The serial order effect—the tendency for later responses to a divergent thinking task to be better than earlier ones—is one of the oldest and most robust findings in modern creativity work. But why do ideas get better? Using new methods that afford a fine-grained look at temporal trajectories, we contrasted two explanations: the classic spreading activation account and a new account based on executive and strategic aspects of creative thought. After completing measures of fluid intelligence and personality, a sample of young adults (n = 133) completed a 10-min unusual uses task. Each response was time-stamped and then rated for creativity by three raters. Multilevel structural equation models estimated the trajectories of creativity and fluency across time and tested if intelligence moderated the effects of time. As in past work, creativity increased sharply with time and flattened slightly by the task's end, and fluency was highest in the task's first minute and then dropped sharply. Intelligence, however, moderated the serial order effect—as intelligence increased, the serial order effect diminished. Taken together, the findings are more consistent with a view that emphasizes executive processes, particularly processes involved in the strategic retrieval and manipulation of knowledge, than the simple spreading of activation to increasingly remote concepts.
Added by: alexb44 |
| Notes |
|
Examines new perspectives for the serial order effect. By looking at the connection between fluid intelligence and the fluidity of idea generation as well as their creative quality, the paper shows evidence it isn't necessary to work through bad ideas to arrive at good ones, and that creative ideas can be recruited and developed without associative spread. Intelligent people don't 'wait around' for the bad ideas to disappear, they are capable of producing good ideas immediately, steered by their fluid intelligence.
Essentially, the serial order effect diminished with higher fluid intelligence, almost to the point it wasn't apparent at all for very intelligent people. The paper isn't dismissing the serial order effect, merely giving it more nuance. Added by: alexb44 Last edited by: alexb44 |
| Quotes |
|
p. 311
"Second, executive switching—when people stop generating ideas from one conceptual category and switch to another— unfolds over time. Research on verbal fluency tasks, which are close cousins of divergent thinking tasks (Carroll, 1993), shows that people higher in intelligence and working memory span switch categories more often (Troyer, Moscovitch, & Winocur, 1997; Unsworth, Spillers, & Brewer, 2011)."
Added by: alexb44
(5/12/26, 5:02 AM)
|
|
p. 311
"Third, interference management during a divergent thinking would lead to later responses being better than earlier ones. Creating good responses involves managing interference from obvious uses, previously generated responses, and responses closely connected to the object’s salient features (e.g., a brick’s color, shape, and density). Managing interference is a central executive process (Unsworth, 2010), and several cognitive abilities, particularly fluid intelligence and working memory span (Burgess, Gray, Conway, & Braver, 2011), predict how well people can do it."
Added by: alexb44
(5/12/26, 5:03 AM)
Comments: It's likely that initial searches for alternative uses in divergent thinking tasks activates highly related associates in semantic memory. This is a type of interference of tightly knit conceptual knowledge domains, and navigating them without defaulting to accepting the salient knowledge requires executive processes and abilities associated with the control of attention. This is interesting in relation to AI. The basis of LLMs is to increasingly strengthen these tightly knit conceptual domains to the point that they become true, which in turn makes it worse and worse at not accepting these domains. Since this is such a fundamental part of creativity, it makes the models increasingly homogeneous. Added by: alexb44 (5/12/26, 5:03 AM) |
|
p. 311
Fluid intelligence
"A straightforward way to examine the role of executive processes is to examine the role of fluid intelligence, a domain-general executive ability (Carroll, 1993). People higher in fluid intelligence can use complex strategies more effectively (Nusbaum & Silvia, 2011a, Study 2), switch idea categories more often (Nusbaum & Silvia, 2011a, Study 1), manage interference (Unsworth, 2010), and discern patterns, rules, and structure in complex problems (McGrew, 2005). Fluid intelligence is closely tied to working memory span (Su¨ß, Oberauer, Wittman, Wilhelm, & Schulze, 2002), another ability fundamental to the deliberate control of cognition and attention. Fluid intelligence can thus represent variability in how well people can exert control over the mental resources needed to generate good ideas"
Added by: alexb44
(5/12/26, 5:03 AM)
Comments: Represents variability in how well people exert control over mental resources when generating good ideas. Includes using complex strategies, switching idea categories often, managing interference and discerning patterns, rules and structures in complex problems. Added by: alexb44 (5/12/26, 5:03 AM) |
|
p. 314
"The serial order effect diminished as intelligence increased, and it was effectively absent at the highest levels of intelligence in our sample (see Figure 2)."
Added by: alexb44
(5/12/26, 5:05 AM)
Comments: The serial order effect in divergent thinking might result from operation of top-down processes and not simply spreading activation. i.e. drawing on experiences, expectations, making sense of things. Relates strongly to the predictive mind. Added by: alexb44 (5/12/26, 5:05 AM) |
|
p. 314
"If anything, the long-standing spatial metaphor of moving away from close, obvious ideas toward remote, original ones may be misleading (cf. Moran, 2009). Good ideas needn’t be “farther away” in semantic space than bad ideas—indeed, good ideas are often generated by people on the spot, not waiting to be found in the frontiers of memory (Gilhooly et al., 2007). Instead, people who are more effective at managing their minds can recruit the strategies and processes needed to generate ideas despite interference generated from obvious ideas and entrenched ways of thinking."
Added by: alexb44
(5/12/26, 5:05 AM)
Comments: suggests that good ideas need not be remote in the semantic space. Fluidly intelligent people can employ effective strategies to minimise interference and deviate from "obvious ideas and entrenched ways of thinking". Added by: alexb44 (5/12/26, 5:05 AM) |
|
p. 315
"Traditional methods for scoring divergent thinking, particularly uniqueness scoring and its variants, lead to unusually high correlations between creativity and fluency (e.g., Hocevar, 1979; Silvia, 2008b). Subjective ratings, in contrast, appear to effectively divorce creative quality from mere quantity (Silvia, Martin, & Nusbaum, 2009; Silvia et al., 2008)."
Added by: alexb44
(5/12/26, 5:06 AM)
Comments: Subjective scoring separates creative quality from fluency. This is interesting re. AI also... Evaluating the ideas reveal the limitations. Added by: alexb44 (5/12/26, 5:06 AM) |
|
p. 316
"If smart people are simply inhibiting the same prepotent associates that everyone else is typing, then we would expect them to take longer to type their first response, assuming that the rate of semantic spread is similar across intelligence levels and that the process of inhibiting and rejecting ideas requires time."
Added by: alexb44
(5/12/26, 5:07 AM)
Comments: Findings differ from spreading activation model. Such a view suggests that everyone works through bad/obvious ideas, but intelligent people don't take longer to begin typing their ideas, so they're not merely waiting for the bad ideas to disappear. The rate of semantic speed isn't the same, it depends on intelligence. Added by: alexb44 (5/12/26, 5:07 AM) |
|
p. 316
"Two other mechanisms—strategy use and goal maintenance— deserve mention. Strategy use received an extensive treatment in Gilhooly et al.’s (2007) protocol analysis of people’s verbal reports during unusual uses tasks. As we mentioned earlier, that research found a wide range of strategies that varied in effectiveness. People who came up with better ideas were able to stop the most intuitive strategy (search memory for known uses), identify a better one, and keep it in mind while enacting it. Goal maintenance—the ability to keep a task’s goal in mind while performing the task— has not yet been studied in creativity tasks, but it is probably important. People with better cognitive control, such as people with high working memory spans, are less likely to have a task’s goal (such as “to be creative” and “generate only creative uses”) slip from mind during the task (Kane & Engle, 2003; Marcovitch, Boseovski, Knapp, & Kane, 2010). "
Added by: alexb44
(5/12/26, 5:09 AM)
Comments: Goal maintenance is under-researched, but potentially really important, according to the authors. This paragraph highlights just how complex the current (2012) understanding of the executive processes that constitute high quality creative work really is. This is also interesting in regards to AI and creative work - they do not themselves have any intrinsic motivation or goals, so for the AI's work to be able to represent that of the user, the prompt needs to contain the goal very specifically as it exists in the head of the user. This seems quite unlikely to me. Added by: alexb44 (5/12/26, 5:09 AM) |