✉ joonop99@snu.ac.kr · snu-csnl.com
Click a card to filter repositories by topic. auto-research is co-developed with @ysys143.
Integrated M.S./Ph.D., Brain and Cognitive Sciences, Seoul National University · 2024.03 – Present
B.S., Convergent Biotechnology and Advanced Materials Science, Kyung Hee University · 2020.03 – 2024.02
I study how uncertainty in the representation of absolute magnitude (e.g., size, time, numerosity) is carried into its relative representation within a context. So far, I have used computational models to predict human behavior and tested those predictions by collecting and fitting behavioral data; next, I plan to extend this work to fMRI and neural network models.
- Magnitude estimation — size, pitch, time, numerosity, value
- Context-dependent normalization[1][2]
- History effects[2][3][4]
- Granularity effect[2]
- Cognitive models
- Bayesian observer models — to explain how humans combine noisy measurements with prior knowledge to produce Bayes-optimal behavior.
- Efficient-coding and rate-distortion models — to explain how perception and behavior are optimized under limited coding capacity.
- Neural models
- Task-optimized RNNs — to explain temporal dynamics, such as the accumulation and drift of working memory, and how neural representations align across contexts.[1][3][5]
- Generative models (e.g., β-VAE) — to obtain both resource-optimal behavior and emergent neural representations from a single model trained under a resource constraint, extending resource-constrained Bayesian observers from process models to neural models.[6][7][8]
- CNNs — to trace the origin of encoding biases to natural image statistics and to study how magnitude representations emerge through learning, using models inspired by the primate ventral visual pathway.
- Lab experiments — psychophysical perceptual decision-making and estimation tasks with human participants, run in a darkened room with a chin rest for controlled viewing.
- Web experiments — large-sample and longitudinal studies run on Prolific or the CSNL web experiment platform.
- Stimuli — rings, random-dot kinematograms (RDKs), rotating bars, dot arrays, and more.
Projects by dimension (🔒 private, available on request)
| Dimension | Projects |
|---|---|
| Size · pitch (granularity effect) | 🔒gran_cann · 🔒granularity-effect-sim |
| Size | 🔒AR_Estimation_n51 · 🔒RingRepSca · 🔒RingRepSca-Bayesian · 🔒RingRepSca_SEM |
| Time | 🔒Time2Dist · 🔒Time2Dist_exante · 🔒Time2Dist_Exp1 · 🔒Time2Dist_Exp1_N12 · 🔒Time2Dist_Exp2 · 🔒Time2Dist_Exp2_Scaling_Trajectory · 🔒SD_time2dist_exp2 · 🔒boec_hmc |
| Numerosity | 🔒Numerosity · 🔒Uncertainty |
| Bayesian observer + efficient coding (BOEC) | 🔒boec_simulator · boec_review · 🔒boec_hmc |
References
- Sheahan, H., Luyckx, F., Nelli, S., Teupe, C., & Summerfield, C. (2021). Neural state space alignment for magnitude generalization in humans and recurrent networks. Neuron, 109(7), 1214–1226.e8. doi
- Lee, H., Lim, J., & Lee, S.-H. (2025). Belief updating in decision-variable space: More fine-grained choices attract future ones more strongly. iScience, 28(7), 112844. doi
- Gu, H., Lee, J., Kim, S., Lim, J., Lee, H.-J., Lee, H., Choe, M. J., Yoo, D., Ryu, J. H., Lim, S., & Lee, S.-H. (2025). Attractor dynamics of working memory explain a concurrent evolution of stimulus-specific and decision-consistent biases in visual estimation. Neuron, 113(20), 3476–3490.e9. doi
- Lim, J., & Lee, S.-H. (2023). Spatial correspondence in relative space regulates serial dependence. Scientific Reports, 13, 18162. doi
- Flesch, T., Juechems, K., Dumbalska, T., Saxe, A., & Summerfield, C. (2022). Orthogonal representations for robust context-dependent task performance in brains and neural networks. Neuron, 110(7), 1258–1270.e11. doi
- D'Amato, L., Lancia, G. L., & Pezzulo, G. (2025). The geometry of efficient codes: How rate-distortion trade-offs distort the latent representations of generative models. PLOS Computational Biology, 21(5), e1012952. doi
- Fang, Z., & Sims, C. R. (2025). Humans learn generalizable representations through efficient coding. Nature Communications, 16, 3989. doi
- D'Amato, L., Nuzzi, D., Testolin, A., Stoianov, I. P., Zorzi, M., & Pezzulo, G. (2025). A rate-distortion perspective on the emergence of number sense in unsupervised generative models. arXiv:2512.19450. arXiv
And, lately, a fair amount of vibe coding.
Seoul National University
| Fall 2026 | Probabilistic Brain (M0000.002100) |
| Spring 2026 | Methods in Brain and Cognitive Sciences (339.502) |
| Fall 2025 | Brain and Cognitive Science of Vision (M1426.004000) |
| Fall 2024 | Data Collection and Statistical Analysis for Functional MRI (M2480.000600) |
| Fall 2024 | Brain and Cognitive Science of Consciousness (M1426.004100) |
Conference poster
Park, J., & Lee, S.-H. (2026). The fate of uncertainty in the transformation from absolute to relative magnitude [Poster presentation]. CBrain 2026.
