A Fan Effect for Creative Thought: Semantic Richness Facilitates Idea Quantity but Constrains Idea Quality

Abstract

Creative thinking has long been associated with spreading of activation through concepts within semantic memory. Despite its theoretical importance, little is yet known about how semantic memory structure facilitates and constrains idea production. We examine one potential influence on spreading activation during divergent thinking known as the fan effect: increasing knowledge about a concept leads to increasing interference from conceptually related information. Specifically, we tested whether cue association size—an index of semantic richness reflecting the average number of elements associated with a given concept—impacts the quantity (fluency) and quality (originality) of responses generated during the alternate uses task (AUT). We hypothesized that low-association AUT cues should benefit originality at the cost of fluency because such cues are embedded within a semantic network with fewer conceptual elements, thus yielding lesser interference from closely-related concepts. This hypothesis was confirmed in three experiments. Furthermore, we found an interaction with individual differences in fluid intelligence in the low-association AUT cues, suggesting that constraints of sparse semantic knowledge can be overcome with top-down intervention. The findings indicate that semantic richness differentially impacts the quality and quantity of generated ideas, and that cognitive control processes can facilitate idea production when conceptual knowledge is limited.

The structure of semantic representation shapes controlled semantic retrieval

“…On the other hand, a combination of low CC, high ASPL, and high Q is characteristic for sparse connectivity network structure that constrains the spreading of activation among concepts and/or semantic clusters (i.e., yields less efficient propagation of information), thus alleviating the load on controlled inhibitory processing imposed by multiple and strongly interconnected associates. In support of this notion, increased number of associates to a concept has recently been considered a source of semantic interference imposing an increased need for cognitive control, especially when the typical/dominant responses are to be suppressed (Beaty et al, 2019). Taken together, these findings indicate that the structural properties of semantic representation may parametrically modulate the engagement of various (controlled) processes supporting semantic memory retrieval, and thus affect higher-level cognition.…”

Automating Creativity Assessment with SemDis: An Open Platform for Computing Semantic Distance

“…Dumas & Dunbar, 2014). Recent work suggests that this low inter-item correlation could be due to variability in item (object) characteristics such as semantic object features (Beaty, Kenett, Hass, & Schacter, 2019) and word frequency (Forthmann et al, 2016). At the same time, low inter-task fluency correlations have not been consistently reported in the literature; for example reported high standardized factor loadings on an AUT fluency latent variable (suggesting strong reliability) and reported inter-task correlations for AUT items ranging from .57 to .71.…”

Automating creativity assessment with SemDis: An open platform for computing semantic distance

“…Dumas & Dunbar, 2014). Recent work suggests that this low inter-item correlation could be due to variability in item (object) characteristics such as semantic object features (Beaty, Kenett, Hass, & Schacter, 2019) and word frequency (Forthmann et al, 2016). At the same time, low inter-task fluency correlations have not been consistently reported in the literature; for example, Jauk et al (2014) reported high standardized factor loadings on an AUT fluency latent variable (suggesting strong reliability) and reported inter-task correlations for AUT items ranging from .57 to .71.…”