Conference PaperPDF Available

Working Memory and Crystallized Knowledge in Visual Analogy

Authors:

Abstract

Developmental, neuropsychological, and computational studies have suggested the importance of both relational knowledge and working memory in analogical reasoning. In this study, we investigated the extent to which individual differences in working memory (WM) and crystallized knowledge (Gc) predicted accuracies on a visual analogy verification task. In the task participants were asked to compare geometric shapes varying in several parametric dimensions (Sweis, Bharani, & Morrison, 2012). Across problems we varied the problem difficulty by factorially manipulating relational complexity and relational distraction. As in many studies of matrix reasoning, both WM and Gc composite measures were reliably correlated with overall visual analogy performance. However, only WM, but not Gc, predicted the effect of relational complexity on visual analogy performance. We believe these results further confirm the importance of working memory as a distinct neurocognitive resource necessary for processing relationally complex analogies.
Both verbal and visual analogical reasoning are frequently positioned at the psychometric center of the construct of fluid intelligence (Snow,
Kyllonen, & Marshalek, 1984), a intelligence construct dependent on working memory and interference control (Unsworth, 2010). !
Real world analogy is also dependent on knowledge, hypothesized to be an important factor in the development of analogical reasoning (e.g.,
Goswami, 2001; Ratterman & Gentner, 1998; Doumas, Morrison, & Richland, 2009; Morrison, Dourmas, & Richland, 2011).!
Working memory and interference control (e.g., Halford, 1993; Waltz et al., 2000; Morrison, Holyoak, & Truong, 2001; Morrison, 2005; Richland,
Morrison, & Holyoak, 2006; Morrison, Doumas, & Richland, 2011; Thibaut, French, & Vezneva, 2010ab; Viskontas et al., 2004) have also been
implicated as important for analogical reasoning in both children and adults.!
Recently Sweis, Bharani, and Morrison (2012) developed a novel visual analogy paradigm that varied task difficulty by manipulating relational
complexity and relational distraction. Importantly working memory span helped to predict the neural correlates of reasoning with low-span
individuals encoding irrelevant relations during processing and thereby requiring greater inhibitory control during final processing.!
In this individual differences study we administered commonly used working memory, interference control, and crystallized knowledge measures
along with our visual analogy paradigm to investigate to what extent these constructs account for unique variance in visual analogy performance.!
175 Wayne State University undergraduates participated in a multi-day testing session. 134
completed all tasks and are used in the analyses reported here.!
A working memory component (WM) was calculated by averaging the z-scores of the automated
Operation and Reading Span (Unsworth, Heitz, Schrock, & Engle, 2005) measures.!
Interference control (IC) was measured using a version of the Brown-Peterson task with category
switches (Unsworth, 2010) to measure sensitivity to proactive interference. A z-score for the
number of memory intrusions was calculated for use in analyses.!
A knowledge component (K) was calculated by averaging the z-scores a general knowledge
(adapted from Unsworth, 2010) and Shipley Vocabulary (Shipley, 1940) measures.!
Participants also completed 144 visual analogy problems (Sweis, Bharani, & Morrison, 2012) of
varying difficulty (see figure to right). D-prime was calculated as a measure of accuracy (VA).!
In this study we present evidence that working memory, interference control, and knowledge all account for unique variance in predicting
performance on a visual analogy task.!
Future efforts should focus on how these abilities interact in the service of analogy and its development in children and older adults. !
VA!
WM!
IC!
K!
VA!
.52!
p<.001!
-.26!
P=.003!
.48!
p<.001!
WM!
.52!
p<.001!
-.14!
p=.12!
.37!
p<.001!
IC!
-.26!
p=.003!
-.14!
p=.12!
-.04!
P=.65!
K!
.48!
p<.001!
.37!
p<.001!
-.04!
p=.65!
-1!
0!
1!
2!
3!
4!
-3!-2!-1!0!1!2!3!
VA d-prime!
WM (standard score)!
-1!
0!
1!
2!
3!
4!
-3!-2!-1!0!1!2!3!
IC (standard score)!
K (standard score)!
ß!
Std
Error!
Std!
ß!
t!
p!
constant!
2.010!
.078!
25.6!
<.001!
WM!
.482!
.096!
.380!
5.0!
<.001!
IC!
-.208!
.080!
-.182!
-2.6!
.011!
K!
.376!
.095!
.296!
3.9!
<.001!
R2!
.372!
adj. R2!
.357!
Examples of several different visual analogy problem types.
Participants were cued to attend to either 1 or 2 relations
present in the stimuli. On some trials non-cued relations
distracted participants from obtaining the correct answer, on
others cued relations did not map correctly (invalid).!
1-Relation
Vali d
No-Distraction
1-Relation
Vali d
Distraction
Invalid
2-Relation
Vali d
Distraction
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.