Kiri Kuroda
LightNews LightNews
kirikuroda.bsky.social
Kiri Kuroda
@kirikuroda.bsky.social
64 followers 96 following 8 posts
Postdoc @arc-mpib.bsky.social, @mpib-berlin.bsky.social Interested in human social/group decision-making. Also working as a social media team at ARC. kirikuroda.com/en
Posts Replies Media Videos
kirikuroda.bsky.social
5/ People adapted to this social situation.

💡🐢Slow people relied less on the fast ones.

💡🐇Fast people relied more on the slower ones and became a bit more cautious.

Our experiment and simulations show that pairs performed as well as individuals. These results suggest flexible use of social-info.
Figure 4. Social drift rate and simulation results. (a) Social drift rates and within-pair differences in thresholds. Fitted lines indicate the mean conditional effects (solid line indicates credible effect; dashed line indicates non-credible effect); shaded areas indicate the 95% uncertainty intervals of the mean. (b) Simulated individual accuracy as a function of social drift rate. Error bars indicate the standard errors of the means. (c) Simulated individual accuracy as a function of within-pair difference in threshold, social drift rate and the lower threshold in a pair. White dots and dashed spline curves indicate the optimal social drift rates (y-axis).
July 16, 2025 at 9:34 AM
kirikuroda.bsky.social
4/ What did we find?

🐇Fast (but less accurate) people often decided first and set the tone.

🐢Slower (but more accurate) people followed.

Social accuracy depended on the faster person, not the better-informed one. So, did the pair's accuracy get worse ...?

👉But here's the twist:
Figure 2. Individual differences in the personal condition. (a) Mean individual response time (RT) and accuracy in the personal condition. Each dot represents one participant. The blue line indicates the locally estimated scatterplot smoothing (LOESS); the shaded area indicates the 95% confidence interval. (b) Participants’ personal drift rates and thresholds in the personal condition. Each dot represents one participant. (c) Predicted individual accuracy. Colours represent predicted individual accuracy from 1000 simulations for each combination of threshold and personal drift rate. See panel b for legend. (d) Predicted mean individual response times. Colours represent the predicted mean individual response time from 1000 simulations. Figure 3. Pairs' mean accuracy and drift–diffusion model (DDM) parameters. (a) Individual accuracy in the personal and social conditions. (b) Accuracy of second decisions in a pair. Each dot indicates a participant but can be based on a different number of underlying trials. (c,d) Probability of being the leader and within-pair difference in (c) threshold and (d) personal drift rate. (e,f ) Leaders’ accuracy as a function of higher and lower (e) thresholds and (f ) personal drift rates in a pair. (g,h) Pairs’ mean accuracy as a function of higher and lower (g) thresholds and (h) personal drift rates in a pair. In panels c–h, fitted lines indicate the mean conditional effects (solid line = credible effect; dashed line = non-credible effect); shaded areas indicate the 95% uncertainty intervals of the mean.
July 16, 2025 at 9:34 AM
kirikuroda.bsky.social
3/ 🧑‍💻To address this, we ran an online study, where people made perceptual judgments while seeing their partner's choice in real time and making use of it.

📈We used drift–diffusion models to capture how competent and cautious each person was and how much they relied on their partner's social info.
Figure 1. Task and drift–diffusion models (DDMs). (a) Trial flow. (b) Illustration of the DDM. (c) Illustration of the social DDM (SDDM) with two individuals. Non-decision time is omitted from panels (b) and (c) for simplicity.
July 16, 2025 at 9:34 AM
kirikuroda.bsky.social
"Individual differences in speed–accuracy trade-off influence social decision-making in dyads"

🚨Our paper has been published in Proceedings of the Royal Society B doi.org/10.1098/rspb...

w/ Alan Tump @alantump.bsky.social, Ralf Kurvers @ralfkurvers.bsky.social

@royalsocietypublishing.org

1/ 🧵👇
Abstract: Speed–accuracy trade-offs are a fundamental aspect of decision-making, requiring individuals to balance collecting more information against making faster decisions. Although speed–accuracy trade-offs have been studied at the individual level, their role in human decision-making in social settings remains poorly understood—even though faster, and possibly more error-prone, decisions often have more social influence than slower decisions. We examined how individual differences in speed–accuracy trade-off preferences shape decision-making in pairs, using an interactive online experiment and drift–diffusion modelling. Participants first performed a perceptual task alone, allowing us to estimate their individual drift rates and decision thresholds, the key cognitive determinants of speed–accuracy trade-off preferences. They then performed the task in pairs, sharing decisions in real-time. Pair accuracy depended on the faster (and thus more error-prone) member, and not on the slower (but more accurate) member. Social decisions were not worse than individual ones because faster members increased their thresholds in the social condition and became more accurate, while slower members incorporated less social information. These findings show that individuals adjusted their social information use to the speed–accuracy trade-off preferences of their partners, highlighting the importance of such individual differences for understanding social behaviour.
July 16, 2025 at 9:34 AM
kirikuroda.bsky.social
How does speed–accuracy trade-off shape pair decisions? Using cognitive models, we show that group accuracy depends on the faster (and more error-prone) members, but that members use social information adaptively.

Preprint: osf.io/preprints/ps...
w/ Alan Tump, Ralf Kurvers @arc-mpib.bsky.social
October 11, 2024 at 11:40 AM