1 of 1

Johan A. Gamba¹, Abigail Stelmashenko¹, Allison Soule¹, Alexei Dawes² , Hakwan Lau², Brian Odegaard¹

¹Department of Psychology, University of Florida,USA, ²Riken Center for Brain Sciences, Wako, Japan

Task-irrelevant motion information modulates working memory representations

Background

Research Questions:

Experimental design:

Perceiving while holding and manipulating information in mind (i.e. working memory - WM) is a common human behavior. However, they are generally studied as isolated processes.

Previously, Kang and colleagues (2011) found that perception of motion stimuli is influenced by previous orientations held in working memory.

We aim to understand how task-irrelevant motion information modulates (1) the presence of perceptual effects (i.e., the Motion Repulsion Effect - MRE; (Marshak and Sekuler, 1979)), and (2) working memory judgments of motion stimuli.

Can we match performance across different dot speeds as participants make judgments about density, speed, & orientation?
Does the MRE replicate while maintaining multiple stimulus features in working memory?

Matched SDT metrics across different dot speeds

The MRE replicates while holding multiple features in mind

Conclusions, future directions & contact info.

Signal Detection Theory (SDT):

No significant differences in HR, FAR, MR,CRR, and mean SDT metrics: (d’, meta d’, M-diff, M-ratio, criterion).

Kang, M.-S., Hong, S. W., Blake, R., & Woodman, G. F. (2011). Visual working memory contaminates perception. Psychonomic Bulletin & Review, 18(5), 860–869. https://doi.org/10.3758/s13423-011-0126-5

Maniscalco, B., & Lau, H. (2012). A signal detection theoretic approach for estimating metacognitive sensitivity from confidence ratings. Conscious Cogn, 21(1), 422–430. https://doi.org/10.1016/j.concog.2011.09.021

Marshak, W., & Sekuler, R. (1979). Mutual Repulsion Between Moving Visual Targets. Science, 205(4413), 1399–1401. https://doi.org/10.1126/science.472756

"Speed" performance decreased in the “Faster” experiment. “Faster” also had fewer "high confidence" correct responses & more "low confidence" incorrect responses. "Slower" experiment had marginally lower overall means.

Task 2: Report the orientation judgment!

Task 3: Select which property changed and how confident are you!

Pattern 1

Pattern 2

Orientation

report

Pattern 3

2AFC Decision

Feedback

Task 1: Keep pattern 1 in your mind, to compare pattern 1 & 3!

400 ms

500 ms

3000 ms

500 ms

unlimited

900 ms

EVERYONE

RDK1 = RDK2

SAME SPEED

RDK1 > RDK2

SLOWER

RDK1 < RDK2

FASTER

N=15

N=16

Training

4 min (24 trials)

Type of Sessions:

Session 1

30 min (200 trials)

Session 2

30 min (200 trials)

VVIQ-2

5 min(32 items)

Staircasing each feature resulted in matched d’ & meta-d’ values across conditions.

The Motion Repulsion effect is robust to task-irrelevant motion information modulation.

SDT → modelling framework to study how to differentiate relevant (signal) from irrelevant (noise) information in uncertain contexts.

Orientation

Dot Density

SLOWER

However, accuracy, staircase values, and confidence proportions significantly differed across features.

Speed

All angular separations involving MRE (-60°,-30°,30°,60°) were significantly different from -90°. No significant effects across experiments. MRE depends on higher proportion of trials.

2. The MRE replicates while maintaining multiple properties in WM through the proportion of repulsed trials rather than the degree of error.

Meta-d’ (Maniscalco et al., 2012): a measure of how well my confidence tracks my decision. M-ratio & M-difference: a measure of how well my confidence tracks a decision at a certain performance level.

N=31

Mean Accuracy

Slower

Same

Faster

Mean accuracy per experiment & trial feature.

76.55

75.15

73.90

Angular Separation between RDK1 and RDK2 (°)

Mean Error Distance (°) & SEM

Error distance as a function of Angular Difference across experiments

Effect:	Coefficient	std. error	p-value	Ef. size
Intercept(-90°)	-0.75	0.663	0.258	-0.227
-60°	-2.30	0.867	0.008*	-0.696
-30°	-4.47	0.867	<0.001*	-1.353
30°	5.98	0.867	<0.001*	1.809
60°	4.87	0.867	<0.001*	1.473
90°	0.92	0.867	0,287	0.279
Exp - Faster	0.35	1.045	0.738	0.106
Exp - Slower	-0.82	1.126	0.465	-0.249
Slower & 90°	2.36	1.559	0.130	0.715
Slower & - 30°	2.04	1.559	0.190	0.618
Slower & 30°	1.45	1.559	0.354	0.438

SAME

FASTER

Proportion of Trials

Angular Separation between RDK1 and RDK2 (°)

Proportion of Motion Repulsion trials across experiments

Trial proportion per experiment, accuracy

and confidence.

Trial proportion

Slower

Same

Faster

Median staircase value

Density

Speed

Orientation

Staircase mean values per experiment and feature.

Linear mixed effects parameters

Cognitive load limits:

TMS inhibition on dlPFC:

Timeline:

modelling both tasks:

References:

Mean score (a.u.)

d’ prime

Meta-d’

M-difference

M-ratio

criterion

Mean SDT metrics across experiments.

Mean values

Mean HR,FAR,MR, CRR across experiments.

Slower

Same

Faster

Orientation (Estimate=0.061, p<0.001) and Speed (Est = 0.056, p=0.002) performances were significantly different in a Mixed-Effects model while “Slower” marginally decrease it (Est=-0.040,p=0.07). “Faster” reduced speed’s performance (Est=-0.068,p=0.024).

Only experiment model was not significant (F=1.796;p=0.169).

An experiment, accuracy, and confidence interaction was found (chi-square: χ2 = 645.11, 𝑝<0.001). Specifically, the “Faster” experiment had significantly less confident correct responses (standardized residual: -3.14) and higher incorrect low confidence responses (12.78).

Staircase values: Wilcoxon-test was significantly different for speed between Same & Faster (Z=23, p=0.018, r= -0.58). No significant changes in other features/experiments.

Thus, mean performance was successfully matched across experiments although not perfectly at a feature level.

SPA

ENG

@ → j.gambasegovia@ufl.edu

→ @alejandro2529 // @BrianOdegaard2

→ @alejogamba.bsky.social

Poster description:

“Hello, everyone! I’m Alejandro Gamba, representing the PAC LAB at the University of Florida. Today, I’m excited to share our findings on how task-irrelevant motion information modulates working memory representations.

Our research primarily explores how perceptual processes interact with working memory—that is, the information we actively hold and manipulate in our minds—in motion stimuli. Specifically, we investigated two main questions:

1) Can performance remain consistent across different moving dot speeds when participants judge aspects like density, speed, and orientation?

2) Does the MRE replicate while maintaining multiple stimulus features in working memory? For those unfamiliar, the MRE is a perceptual phenomenon where two angles under 60 degrees appear more separated than they actually are.

In our experiment, participants completed a match-to-sample test involving three patterns of dots. They remembered three properties of the first pattern—orientation, speed, and density—reported the orientation of the second, and selected which property changed between the first and the third pattern, while also rating their confidence in these decisions. Notably, we varied the speed of the second pattern of dots as being slower, the same, or faster than the first, to examine its influence on performance.

Our analysis, grounded in Signal Detection Theory, revealed no significant differences in overall performance or metacognitive parameters across the different speeds. However, when we delved into the data by category, we found significant differences in mean performance. The 'Slower' condition marginally reduced overall performance, while the 'Faster' condition notably impacted performance on speed tasks, and significantly influenced both low-confidence incorrect responses and high-confidence correct responses.

From a perceptual standpoint, we successfully replicated the MRE. The effect was most pronounced at angular separations of 30 degrees, and although diminished, it remained significant at 60 degrees—across all experimental conditions. Interestingly, the MRE was explained due to an increased number of trials exaggerating the angular error rather than a sheer increase in error magnitude.

In summary, while we could maintain average performance across different dot speeds, the effects at the feature level were more nuanced. We were also able to replicate the MRE through a higher proportion of repulsed trials rather than by increasing the error degree itself. Our ongoing work includes using Transcranial Magnetic Stimulation (TMS) to inhibit brain regions associated with these processes and explicitly assessing the cognitive load's impact on perceptual effects.

If you have any questions or would like to discuss this further, please feel free to reach out. My contact details are available on the poster. Thank you for your attention, and I hope you enjoyed the poster.”