MEMORY SUBSTITUTION
Supplementary Material for:
When past is present:
Substitutions of long-term memory for sensory evidence in perceptual judgments
Judith E. Fan, J. Benjamin Hutchinson, & Nicholas B. Turk-Browne
Supplemental Figure 1: Model Comparison Results. (a) On visible test trials, the three-component mixture
model (“swap”) provided the best overall fit to the data across experiments, relative to reduced models containing
only the original-color component or only the current-color component, plus the uniform component. Differences
in AIC are plotted relative to the null model containing only the uniform component. The blend model, while
performing well, did not exceed the swap model in any experiment. b) On invisible test trials, where there was no
current color, the two-component model containing the original-color and uniform components significantly
outperformed the null model. The performance of the swap model was comparable, except for the penalty due to
the additional parameter. Error bars indicate 68% confidence intervals.
1
MEMORY SUBSTITUTION
Supplemental Figure 2: Model fits and residuals for each experiment and condition. The mixture modeling
approach used throughout these studies assumes that a report follows either a von Mises distribution or a uniform
guessing distribution (Zhang & Luck, 2008). However, recent studies of visual short-term memory (van den Berg et
al., 2012; Fougnie, Suchow, & Alvarez, 2012; van den Berg, Awh, and Ma, 2014) have found that this basic mixture
model leaves a structured residual: the observed response distribution tends to be ‘peakier’ than the model fit. In
these cases, an alternative ‘variable-precision’ model, in which precision is not fixed but can take on values in a
continuum, has been found to provide a better fit. In the current study of interactions between visual long-term
memory and current sensory information, however, there was not strong evidence for ‘peakiness’ in the data relative
to the model fit. In all panels, dotted lines represent 95% confidence bounds based on simulated data from model.
2
MEMORY SUBSTITUTION
References
Fougnie, D., Suchow, J. W., & Alvarez, G. A. (2012). Variability in the quality of visual working memory.
Nature Communications, 3, 1229.
van den Berg, R., Shin, H., Chou, W.-C., George, R., & Ma, W. J. (2012). Variability in encoding precision
accounts for visual short-term memory limitations. Proceedings of the National Academy of Sciences of the
United States of America, 109(22), 8780–8785.
van den Berg, R., Awh, E., & Ma, W. J. (2014). Factorial comparison of working memory models.
Psychological Review, 121(1), 124–149.
Zhang, W., & Luck, S. J. (2008). Discrete fixed-resolution representations in visual working memory.
Nature, 453(7192), 233–235.
3