Hsin-Hung Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Hsin-Hung Li has a total of 31 co-authors. Here are the top co-authors by number of shared publications:
- Clayton E. Curtis
Department of Psychology, New York University
6 shared publications Marisa Carrasco
Department of Psychology, New York University, New York, New York, USA
6 shared publications- Marisa Carrasco
New York University
5 shared publications - Chien-Chung Chen
Department of Psychology, National Taiwan University, Taipei, Taiwan
4 shared publications Jasmine Pan
New York University
4 shared publications- Wei Ji Ma
New York University
4 shared publications
Latest Publications
- 10 September 2026
- Published by Elsevier in Current Biology
- Vol. 36 (19) , 4943-4952.e4
Distinct oscillatory neural rhythms support representational precision and confidence in working memory
- 10 September 2026
- Published by Elsevier in Current Biology
- Vol. 36 (19) , 4943-4952.e4
Distinct oscillatory neural rhythms support representational precision and confidence in working memory
- Open Access
Probabilistic neural code supports metacognition in value-based decisions in the human brain
Subjective value underlies many everyday decisions, yet the neural representations supporting these valuations are inherently noisy and inevitably subject to uncertainty. Whether humans have metacognitive access to this uncertainty, and how such access is supported neurally, remains unclear. Here, we provide converging behavioral and neural evidence that subjective value is represented probabilistically in the human brain. Behaviorally, participants' explicit uncertainty reports tracked trial-by-trial fluctuations in valuation variability. Neurally, we decoded full probability distributions over value from fMRI population activity in ventromedial prefrontal cortex (vmPFC) and superior parietal lobe and intraparietal sulcus (SPL-IPS). Decoded neural uncertainty predicted behavioral reported uncertainty at the trial, item, and individual levels. These findings indicate that neural populations jointly encode value and uncertainty, extending probabilistic population coding from perception and working memory to subjective value, and providing a neural basis for metacognitive access to uncertainty in value-based decisions.
Open AccessProbabilistic neural code supports metacognition in value-based decisions in the human brain
Subjective value underlies many everyday decisions, yet the neural representations supporting these valuations are inherently noisy and inevitably subject to uncertainty. Whether humans have metacognitive access to this uncertainty, and how such access is supported neurally, remains unclear. Here, we provide converging behavioral and neural evidence that subjective value is represented probabilistically in the human brain. Behaviorally, participants' explicit uncertainty reports tracked trial-by-trial fluctuations in valuation variability. Neurally, we decoded full probability distributions over value from fMRI population activity in ventromedial prefrontal cortex (vmPFC) and superior parietal lobe and intraparietal sulcus (SPL-IPS). Decoded neural uncertainty predicted behavioral reported uncertainty at the trial, item, and individual levels. These findings indicate that neural populations jointly encode value and uncertainty, extending probabilistic population coding from perception and working memory to subjective value, and providing a neural basis for metacognitive access to uncertainty in value-based decisions.
- Open Access
Multiple Oscillatory Neural Rhythms Support Metacognitive Access of Working Memory
Summary: Working memory enables the maintenance of information to guide behavior, yet its representations are inherently noisy and variable. To act effectively, observers must track the uncertainty of their own memory. However, the neural mechanisms supporting uncertainty representations in working memory remain poorly understood, particularly at the level of neural dynamics. Here, we combined electroencephalography (EEG) with a spatial visual working memory task that elicited trial-by-trial uncertainty reports to investigate how oscillatory activity encodes memory uncertainty. We identified two forms of uncertainty supported by distinct oscillatory neural activity. Using a probabilistic encoding–decoding model, we found that memory content can be reconstructed from alpha-band activity, and that the precision of these representations predicts subsequent uncertainty reports, consistent with probabilistic mnemonic representations. In parallel, beta-band activity contained a scalar, magnitude-based, representation of uncertainty. This signal exhibited slow, persistent dynamics across task epochs and tracked uncertainty reports for both the current and previous trials. Together, these findings show that the human brain multiplexes two neural codes for memory uncertainty, representational precision and a scalar confidence signal, across distinct oscillatory rhythms, providing a neural dynamical account of metacognitive access to working memory.
Open AccessMultiple Oscillatory Neural Rhythms Support Metacognitive Access of Working Memory
Summary: Working memory enables the maintenance of information to guide behavior, yet its representations are inherently noisy and variable. To act effectively, observers must track the uncertainty of their own memory. However, the neural mechanisms supporting uncertainty representations in working memory remain poorly understood, particularly at the level of neural dynamics. Here, we combined electroencephalography (EEG) with a spatial visual working memory task that elicited trial-by-trial uncertainty reports to investigate how oscillatory activity encodes memory uncertainty. We identified two forms of uncertainty supported by distinct oscillatory neural activity. Using a probabilistic encoding–decoding model, we found that memory content can be reconstructed from alpha-band activity, and that the precision of these representations predicts subsequent uncertainty reports, consistent with probabilistic mnemonic representations. In parallel, beta-band activity contained a scalar, magnitude-based, representation of uncertainty. This signal exhibited slow, persistent dynamics across task epochs and tracked uncertainty reports for both the current and previous trials. Together, these findings show that the human brain multiplexes two neural codes for memory uncertainty, representational precision and a scalar confidence signal, across distinct oscillatory rhythms, providing a neural dynamical account of metacognitive access to working memory.
- Open Access
- 23 January 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (4)
Dynamics of working memory drift and information flow across the cortical hierarchy
Working memory is supported by widespread and distributed brain regions spanning across the cortical hierarchy. However, how working memory content evolves and is transmitted across cortical regions remains largely unknown. Here, we investigated the flow of working memory information across the cortex using time-resolved fMRI decoding. Across multiple regions in visual and parietal cortex, we found that decoded working memory content drifted over time from the memoranda toward the later errors in memory reports, consistent with the dynamics of memory drift predicted by attractor models. These behaviorally predictive neural memory errors emerged earliest in higher-order dorsal visual area (V3AB) and intraparietal sulcus (IPS0) and then later in early visual cortex (V1), suggesting a propagation of mnemonic information originated in higher-level visual area. Interareal correlation analyses revealed that during memory maintenance, information flowed predominantly in a top–down manner—from higher to lower visual areas, whereas during passive viewing, feedforward dynamics prevailed. Together, our findings demonstrate that working memory maintenance involves temporally structured drift dynamics and feedback-dominated information flow across the cortical hierarchy, providing a mechanistic link between neural population dynamics and the formation of memory errors.
- 23 January 2026
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 123 (4)
Open AccessDynamics of working memory drift and information flow across the cortical hierarchy
Working memory is supported by widespread and distributed brain regions spanning across the cortical hierarchy. However, how working memory content evolves and is transmitted across cortical regions remains largely unknown. Here, we investigated the flow of working memory information across the cortex using time-resolved fMRI decoding. Across multiple regions in visual and parietal cortex, we found that decoded working memory content drifted over time from the memoranda toward the later errors in memory reports, consistent with the dynamics of memory drift predicted by attractor models. These behaviorally predictive neural memory errors emerged earliest in higher-order dorsal visual area (V3AB) and intraparietal sulcus (IPS0) and then later in early visual cortex (V1), suggesting a propagation of mnemonic information originated in higher-level visual area. Interareal correlation analyses revealed that during memory maintenance, information flowed predominantly in a top–down manner—from higher to lower visual areas, whereas during passive viewing, feedforward dynamics prevailed. Together, our findings demonstrate that working memory maintenance involves temporally structured drift dynamics and feedback-dominated information flow across the cortical hierarchy, providing a mechanistic link between neural population dynamics and the formation of memory errors.
- Open Access
Attention induced perceptual traveling waves in binocular rivalry
Cortical traveling waves –smooth changes of phase over time across the cortical surface– have been proposed to modulate perception periodically as they travel through retinotopic cortex. Yet, little is known about the underlying computational principles. Here, we make use of binocular rivalry, a perceptual phenomenon in which perceptual (illusory) waves are perceived when a shift in dominance occurs between two rival images. First, we assessed these perceptual waves using psychophysics. Participants viewed a stimulus restricted to an annulus around fixation, with orthogonal orientations presented to each eye. The stimulus presented to one eye was of higher contrast thus generating perceptual dominance. When a patch of higher contrast was flashed briefly at one position in the other eye, it created a change in dominance that started at that location of the flash and expanded progressively, like a wave, as the previously suppressed stimulus became dominant. We found that the duration of the perceptual propagation increased with both distance traveled and eccentricity of the annulus. Diverting attention away from the annulus reduced drastically the occurrence and the speed of the wave. Second, we developed a computational model of traveling waves in which competition between the neural representations of the two stimuli is driven by both attentional modulation and mutual inhibition. We found that the model captured the key features of wave propagation dynamics. Together, these findings provide new insights into the functional relevance of cortical traveling waves and offer a framework for further experimental investigation into their role in perception.
Open AccessAttention induced perceptual traveling waves in binocular rivalry
Cortical traveling waves –smooth changes of phase over time across the cortical surface– have been proposed to modulate perception periodically as they travel through retinotopic cortex. Yet, little is known about the underlying computational principles. Here, we make use of binocular rivalry, a perceptual phenomenon in which perceptual (illusory) waves are perceived when a shift in dominance occurs between two rival images. First, we assessed these perceptual waves using psychophysics. Participants viewed a stimulus restricted to an annulus around fixation, with orthogonal orientations presented to each eye. The stimulus presented to one eye was of higher contrast thus generating perceptual dominance. When a patch of higher contrast was flashed briefly at one position in the other eye, it created a change in dominance that started at that location of the flash and expanded progressively, like a wave, as the previously suppressed stimulus became dominant. We found that the duration of the perceptual propagation increased with both distance traveled and eccentricity of the annulus. Diverting attention away from the annulus reduced drastically the occurrence and the speed of the wave. Second, we developed a computational model of traveling waves in which competition between the neural representations of the two stimuli is driven by both attentional modulation and mutual inhibition. We found that the model captured the key features of wave propagation dynamics. Together, these findings provide new insights into the functional relevance of cortical traveling waves and offer a framework for further experimental investigation into their role in perception.
- Open Access
- 11 April 2025
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 11 (15) , eadr8015
Neural mechanisms of resource allocation in working memory
To mitigate capacity limits of working memory, people allocate resources according to an item’s relevance. However, the neural mechanisms supporting such a critical operation remain unknown. Here, we developed computational neuroimaging methods to decode and demix neural responses associated with multiple items in working memory with different priorities. In striate and extrastriate cortex, the gain of neural responses tracked the priority of memoranda. We decoded higher-priority memoranda with smaller error and lower uncertainty. Moreover, these neural differences predicted behavioral differences in memory prioritization between and within participants. Trial-wise variability in the magnitude of delay activity in the frontal cortex predicted differences in decoded precision between low- and high-priority items in visual cortex. These results support a model in which feedback signals broadcast from frontal cortex sculpt the gain of memory representations in the visual cortex according to behavioral relevance, thus identifying a neural mechanism for resource allocation.
- 11 April 2025
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 11 (15) , eadr8015
Open AccessNeural mechanisms of resource allocation in working memory
To mitigate capacity limits of working memory, people allocate resources according to an item’s relevance. However, the neural mechanisms supporting such a critical operation remain unknown. Here, we developed computational neuroimaging methods to decode and demix neural responses associated with multiple items in working memory with different priorities. In striate and extrastriate cortex, the gain of neural responses tracked the priority of memoranda. We decoded higher-priority memoranda with smaller error and lower uncertainty. Moreover, these neural differences predicted behavioral differences in memory prioritization between and within participants. Trial-wise variability in the magnitude of delay activity in the frontal cortex predicted differences in decoded precision between low- and high-priority items in visual cortex. These results support a model in which feedback signals broadcast from frontal cortex sculpt the gain of memory representations in the visual cortex according to behavioral relevance, thus identifying a neural mechanism for resource allocation.
- Open Access
Neural mechanisms of resource allocation in working memory
To mitigate capacity limits of working memory, people allocate resources according to an item’s relevance. However, the neural mechanisms supporting such a critical operation remain unknown. Here, we developed computational neuroimaging methods to decode and demix neural responses associated with multiple items in working memory with different priorities. In striate and extrastriate cortex, the gain of neural responses tracked the priority of memoranda. Higher-priority memoranda were decoded with smaller error and lower uncertainty. Moreover, these neural differences predicted behavioral differences in memory prioritization. Remarkably, trialwise variability in the magnitude of delay activity in frontal cortex predicted differences in decoded precision between low and high-priority items in visual cortex. These results suggest a model in which feedback signals broadcast from frontal cortex sculpt the gain of memory representations in visual cortex according to behavioral relevance, thus, identifying a neural mechanism for resource allocation.
Open AccessNeural mechanisms of resource allocation in working memory
To mitigate capacity limits of working memory, people allocate resources according to an item’s relevance. However, the neural mechanisms supporting such a critical operation remain unknown. Here, we developed computational neuroimaging methods to decode and demix neural responses associated with multiple items in working memory with different priorities. In striate and extrastriate cortex, the gain of neural responses tracked the priority of memoranda. Higher-priority memoranda were decoded with smaller error and lower uncertainty. Moreover, these neural differences predicted behavioral differences in memory prioritization. Remarkably, trialwise variability in the magnitude of delay activity in frontal cortex predicted differences in decoded precision between low and high-priority items in visual cortex. These results suggest a model in which feedback signals broadcast from frontal cortex sculpt the gain of memory representations in visual cortex according to behavioral relevance, thus, identifying a neural mechanism for resource allocation.
- Open Access
Neural population dynamics of human working memory
Working memory (WM) depends on the activity of neural populations to maintain internal representations of information that are no longer available in the environment. The neural population dynamics that support WM in the human cortex remain largely unexplored. Utilizing fMRI and time-resolved multivariate analyses, we found evidence of both stable and dynamic WM representations during a classic memory-guided saccade task. Using dimensionality reduction techniques, we identified robust neural subspaces in early visual, parietal and even frontal cortex that were stable throughout the delay period with memorized locations organized within a two-dimensional representation of visual space. The dynamic subspace, which was largely restricted to early visual cortex, had both an early and late delay component during which the WM representation changed. Leveraging models of the receptive fields of voxels comprising each of the visual field maps, we visualized how the dynamic population code changed over time. Early in the trial, the dynamic code in V1 was dominated by a narrowly tuned activation among voxels with receptive fields centered on the location of the peripheral memory target. Late in the trial, the dynamic code was driven by diffuse activation among voxels with receptive fields along a line between the fovea and the peripheral location of the target. Together, the timing, shape, and spatial distribution of the WM dynamics in early visual cortex suggest that the early component encodes feedforward visual inputs, while the late component encodes abstracted task-relevant mnemonic representations resulting from feedback signals from higher-level brain areas. Significance statement Recent empirical evidence of neural dynamics during working memory (WM) challenges established canonical theories of how WM representations are stored in the brain. Utilizing fMRI and time-resolved multivariate analyses, we identified both stable WM codes in visual and association cortex, as well as changing, dynamic codes in early visual cortex. Each of these population codes were composed of low-dimensional topological representations of visual space. Characteristics of the WM dynamics in early visual cortex suggest that feedback signals from higher-brain areas recode sensory inputs into a format matching the demands of the memory-guided task.
Open AccessNeural population dynamics of human working memory
Working memory (WM) depends on the activity of neural populations to maintain internal representations of information that are no longer available in the environment. The neural population dynamics that support WM in the human cortex remain largely unexplored. Utilizing fMRI and time-resolved multivariate analyses, we found evidence of both stable and dynamic WM representations during a classic memory-guided saccade task. Using dimensionality reduction techniques, we identified robust neural subspaces in early visual, parietal and even frontal cortex that were stable throughout the delay period with memorized locations organized within a two-dimensional representation of visual space. The dynamic subspace, which was largely restricted to early visual cortex, had both an early and late delay component during which the WM representation changed. Leveraging models of the receptive fields of voxels comprising each of the visual field maps, we visualized how the dynamic population code changed over time. Early in the trial, the dynamic code in V1 was dominated by a narrowly tuned activation among voxels with receptive fields centered on the location of the peripheral memory target. Late in the trial, the dynamic code was driven by diffuse activation among voxels with receptive fields along a line between the fovea and the peripheral location of the target. Together, the timing, shape, and spatial distribution of the WM dynamics in early visual cortex suggest that the early component encodes feedforward visual inputs, while the late component encodes abstracted task-relevant mnemonic representations resulting from feedback signals from higher-level brain areas. Significance statement Recent empirical evidence of neural dynamics during working memory (WM) challenges established canonical theories of how WM representations are stored in the brain. Utilizing fMRI and time-resolved multivariate analyses, we identified both stable WM codes in visual and association cortex, as well as changing, dynamic codes in early visual cortex. Each of these population codes were composed of low-dimensional topological representations of visual space. Characteristics of the WM dynamics in early visual cortex suggest that feedback signals from higher-brain areas recode sensory inputs into a format matching the demands of the memory-guided task.
- 4 June 2021
- Published by Elsevier in Trends in Neurosciences
- Vol. 44 (8) , 669-686
To look or not to look: dissociating presaccadic and covert spatial attention
- 4 June 2021
- Published by Elsevier in Trends in Neurosciences
- Vol. 44 (8) , 669-686
To look or not to look: dissociating presaccadic and covert spatial attention