Adaptive Memory Distortions Are Predicted by Feature Representations in Parietal Cortex

Abstract

Similarity between memories is a primary cause of interference and forgetting. Exaggerating subtle differences between memories is therefore a potential mechanism for reducing interference. Here, we report a human fMRI study (n = 29, 19 female) that tested whether behavioral and neural expressions of memories are adaptively distorted to reduce interference. Participants learned and repeatedly retrieved object images, some of which were identical except for subtle color differences. Behavioral measures of color memory revealed exaggeration of differences between similar objects. Importantly, greater memory exaggeration was associated with lower memory interference. fMRI pattern analyses revealed that color information in parietal cortex was stronger during memory recall when color information was critical for discriminating competing memories. Moreover, greater representational distance between competing memories in parietal cortex predicted greater color memory exaggeration and lower memory interference. Together, these findings reveal that competition between memories induces adaptive, feature-specific distortions in parietal representations and corresponding behavioral expressions.SIGNIFICANCE STATEMENT Similarity between memories is a primary cause of interference and forgetting. Here, we show that, when remembering highly similar objects, subtle differences in the features of these objects are exaggerated in memory to reduce interference. These memory distortions are reflected in, and predicted by, overlap of activity patterns in lateral parietal cortex. These findings provide unique insight into how memory interference is resolved and specifically implicate lateral parietal cortex in representing feature-specific memory distortions.

Keywords: episodic memory; fMRI; interference; pattern similarity; repulsion.

Copyright © 2021 the authors.

Figures

Figure 1.

Experimental design and procedure. a, Overview of paradigm. On day 1, participants completed 14 Study and Associative Memory Test rounds. During Study, participants were shown object-face pairs; and during Associative Memory Test, participants were shown an object and selected the corresponding face from a set of four choices. The set of four choices included the target face along with the face associated with the object's pairmate. On day 2, participants completed four additional Study and Associative Memory Test rounds before entering the fMRI scanner. During scanning, participants completed a Cued Recall task during which face images were shown and participants recalled the corresponding image and indicated, by button press, the vividness of their recall. After exiting the scanner, participants completed a Color Memory Test during which a face image was shown alongside a grayscale version of the corresponding object. Participants used a continuous color wheel to indicate their memory for the object's color. Finally, participants completed 2 more Associative Memory Test rounds. b, Sample structure of object stimuli. For both the competitive and noncompetitive conditions, pairmate stimuli were 24 degrees apart in color space. For the competitive condition, pairmates were from the same object category; for the noncompetitive condition, pairmates were from distinct categories. For both conditions, some objects had identical colors (Same-color). fMRI pattern similarity for Pairmate and Same-color comparisons were compared against a Baseline comparison of stimuli that were from different object categories and 24 degrees apart in color space. c, d, Responses on the color memory test were used to categorize memory for each object's color as being biased toward or away from the color of the competing object (c) and to measure the signed distance, in degrees, between participants' responses and the true color of the target (d).

Figure 2.

Behavioral results. a, Associative memory performance across the experiment. The overall error rate (pairmate error + other error) was higher in the competitive condition than in the noncompetitive condition for each of the associative memory test sessions (day 1, day 2 prescan, day 2 postscan [not shown]; all p values < 0.0001). Subsequent analyses focused on associative memory performance from the day 2 prescan session. For the day 2 prescan session, participants were significantly more likely to select faces that were associated with the pairmate image (pairmate error) in the competitive condition (6.0% ± 6.6%, mean ± SD) compared with the noncompetitive condition (0.2% ± 0.6%; p < 0.0001), confirming that similarity between pairmates was a source of interference. b, Signed distance of responses in the color memory test. For the competitive condition, mean signed distance was significantly > 0 (p = 0.000003), reflecting a bias away from the color of the pairmate object (repulsion). Signed distance did not differ from 0 in the noncompetitive condition (p = 0.771). The difference between the competitive and noncompetitive conditions was also significant (p = 0.007). c, Percentage of away responses in the color memory test. The percentage of color memory responses “away from” the color of the pairmate object was significantly > 50% for the competitive condition (p = 0.0001), but not for the noncompetitive condition (p = 0.189). The difference between the competitive and noncompetitive conditions was also significant (p = 0.001). d, Relationship between associative memory accuracy and mean signed color memory distance. For the competitive condition, participants with greater mean signed color memory distance (greater repulsion) exhibited better associative memory accuracy [r = 0.50, p = 0.007, one outlier (red dot) excluded for associative memory performance < 3 SDs below mean]. Colored dots represent data from individual participants. Error bars indicate ± SEM. ***p < 0.001. **p < 0.01.

Figure 3.

Neural feature representations as a function of memory competition. a, Anatomical ROIs visualized on the Freesurfer average cortical surface. b, Color information as a function of memory competition. Color information was defined as the fMRI pattern similarity between pairs of same-color objects relative to pattern similarity between baseline pairs of objects (Fig. 1b). Color information was significantly stronger in the competitive than noncompetitive condition (i.e., values > 0) across the set of ROIs as a whole and in LO and vIPS individually (p values < 0.05). c, Pairmate similarity as a function of memory competition. Pairmate similarity was defined as the fMRI pattern similarity between pairmate objects relative to pattern similarity between baseline pairs of objects. Only vIPS showed significantly greater pairmate similarity in the competitive than noncompetitive conditions (p = 0.004). Error bars indicate ± SEM. **p < 0.01. *p < 0.05.

Figure 4.

Neural measures of pairmate (dis)similarity predict color memory bias in vIPS. a, Mean correlation between vIPS pairmate dissimilarity during recall and mean signed color memory distance. Correlations were performed within participant and correlation coefficients were z-transformed. For the competitive condition, the mean correlation was significantly positive (p = 0.004), indicating that greater pairmate dissimilarity in vIPS was associated with a stronger bias to remember pairmates' colors as away from each other. There was no correlation between vIPS pairmate dissimilarity and signed color memory distance for the noncompetitive condition (p = 0.566). b, Relationship between vIPS pairmate dissimilarity (binned into low, medium, high groups) and mean signed color memory distance (purple) and associative memory accuracy (teal). Mean signed color memory distance and associative memory accuracy each significantly varied as a function of vIPS dissimilarity (p values <0.05), with greater vIPS dissimilarity associated with greater mean signed color memory distance and higher associative memory accuracy. ***p < 0.001. *p < 0.05.