Sixing Chen

Sixing Chen

@sixingchen.bsky.social

PhD student at NYU

91 followers240 following19 posts

Sixing Chen reposted

Tianyuan Teng
@tyteng.bsky.social

You'd think people prefer the simplest explanation of an uncertain world — Occam's razor. Our data says no, across 2 modalities, 2 tasks, 8 experiments: people often perceive illusory structure that isn't there and prefer moderate complexity. Out now in Nature Comms! www.nature.com/articles/s41...

nature.comHuman learning of probability distributions is biased toward moderate structural complexity - Nature CommunicationsHumans build internal models from online observations to adapt to new environments. Here, the authors show that individuals are biased towards building models with moderate structural complexity, rega...

Sixing Chen reposted

Moufan Li
@moufanli.bsky.social

Our paper is out in @natmachintell.nature.com! We trained multiple RNNs to perform free recall. The best-performing ones learned a strategy akin to the memory palace technique. See thread below for more info. www.nature.com/articles/s42...

nature.comA neural network model of free recall learns multiple memory strategies - Nature Machine IntelligenceLi et al. show that recurrent neural networks optimized for free recall discover diverse, human-like memory strategies beyond classical temporal context models, with top models using an index-based me...

Sixing Chen reposted

Jonathan Nicholas
@jonathannicholas.bsky.social

We make flexible choices in new situations by knitting together information from separate relevant memories. But what governs which memories are retrieved and when? In a new preprint, we captured how people build decision variables from different memories by tracking their gaze on a blank screen.

biorxiv.orgFlexible decisions arise from resource-rational memory samplingFlexible decision making depends on retrieving and recombining memories. Yet because this process unfolds covertly, its governing principles remain unknown. Here we use gaze reinstatement to uncover t...
Sixing Chen
@sixingchen.bsky.social

New preprint w/ @fredcallaway.bsky.social! How does the brain decide which computations to run? We combine rational meta-reasoning with a meta-learning algorithm to build a recurrent network that learns to select computations. www.biorxiv.org/content/10.6...

biorxiv.orgLearning to select computations in recurrent neural circuitsTwo hallmarks of biological computation are its flexibility and efficiency. These features are often attributed to cognitive control processes that balance external utility against computational cost. However, how the brain could implement such adaptive control remains unknown. Here, we provide one possible answer by combining the computational theory of rational meta-reasoning with a meta-learning algorithm recently proposed as a model of prefrontal cortex. This yields a recurrent neural network model that learns to select computations. In simple choice tasks, the model approximates the algorithms and representations of optimal symbolic models and reproduces neural dynamics observed in macaque orbitofrontal cortex. In multi-step planning tasks, the model replicates key behavioral signatures of human planning strategies and captures human neural dynamics associated with step-by-step mental simulation. Our framework unifies meta-reasoning and meta-learning by showing that learning to reason can be understood as learning to learn from information generated by one’s own cognitive operations, providing a mechanistic account of how adaptive control of thought can be implemented in neural systems. ### Competing Interest Statement The authors have declared no competing interest.
View full thread
Sixing Chen
@sixingchen.bsky.social

Zooming out: our work unifies meta-reasoning and meta-learning by showing that learning to reason can be understood as learning to learn from information generated by one’s own cognitive operations. Check out our preprint for more information!

Sixing Chen
@sixingchen.bsky.social

New preprint w/ @fredcallaway.bsky.social! How does the brain decide which computations to run? We combine rational meta-reasoning with a meta-learning algorithm to build a recurrent network that learns to select computations. www.biorxiv.org/content/10.6...

biorxiv.orgLearning to select computations in recurrent neural circuitsTwo hallmarks of biological computation are its flexibility and efficiency. These features are often attributed to cognitive control processes that balance external utility against computational cost. However, how the brain could implement such adaptive control remains unknown. Here, we provide one possible answer by combining the computational theory of rational meta-reasoning with a meta-learning algorithm recently proposed as a model of prefrontal cortex. This yields a recurrent neural network model that learns to select computations. In simple choice tasks, the model approximates the algorithms and representations of optimal symbolic models and reproduces neural dynamics observed in macaque orbitofrontal cortex. In multi-step planning tasks, the model replicates key behavioral signatures of human planning strategies and captures human neural dynamics associated with step-by-step mental simulation. Our framework unifies meta-reasoning and meta-learning by showing that learning to reason can be understood as learning to learn from information generated by one’s own cognitive operations, providing a mechanistic account of how adaptive control of thought can be implemented in neural systems. ### Competing Interest Statement The authors have declared no competing interest.
View full thread
Sixing Chen
@sixingchen.bsky.social

New preprint w/ @fredcallaway.bsky.social! How does the brain decide which computations to run? We combine rational meta-reasoning with a meta-learning algorithm to build a recurrent network that learns to select computations. www.biorxiv.org/content/10.6...

biorxiv.orgLearning to select computations in recurrent neural circuitsTwo hallmarks of biological computation are its flexibility and efficiency. These features are often attributed to cognitive control processes that balance external utility against computational cost. However, how the brain could implement such adaptive control remains unknown. Here, we provide one possible answer by combining the computational theory of rational meta-reasoning with a meta-learning algorithm recently proposed as a model of prefrontal cortex. This yields a recurrent neural network model that learns to select computations. In simple choice tasks, the model approximates the algorithms and representations of optimal symbolic models and reproduces neural dynamics observed in macaque orbitofrontal cortex. In multi-step planning tasks, the model replicates key behavioral signatures of human planning strategies and captures human neural dynamics associated with step-by-step mental simulation. Our framework unifies meta-reasoning and meta-learning by showing that learning to reason can be understood as learning to learn from information generated by one’s own cognitive operations, providing a mechanistic account of how adaptive control of thought can be implemented in neural systems. ### Competing Interest Statement The authors have declared no competing interest.
View full thread
Sixing Chen
@sixingchen.bsky.social

New preprint w/ @fredcallaway.bsky.social! How does the brain decide which computations to run? We combine rational meta-reasoning with a meta-learning algorithm to build a recurrent network that learns to select computations. www.biorxiv.org/content/10.6...

biorxiv.orgLearning to select computations in recurrent neural circuitsTwo hallmarks of biological computation are its flexibility and efficiency. These features are often attributed to cognitive control processes that balance external utility against computational cost. However, how the brain could implement such adaptive control remains unknown. Here, we provide one possible answer by combining the computational theory of rational meta-reasoning with a meta-learning algorithm recently proposed as a model of prefrontal cortex. This yields a recurrent neural network model that learns to select computations. In simple choice tasks, the model approximates the algorithms and representations of optimal symbolic models and reproduces neural dynamics observed in macaque orbitofrontal cortex. In multi-step planning tasks, the model replicates key behavioral signatures of human planning strategies and captures human neural dynamics associated with step-by-step mental simulation. Our framework unifies meta-reasoning and meta-learning by showing that learning to reason can be understood as learning to learn from information generated by one’s own cognitive operations, providing a mechanistic account of how adaptive control of thought can be implemented in neural systems. ### Competing Interest Statement The authors have declared no competing interest.
View full thread
Sixing Chen
@sixingchen.bsky.social

Meta-reasoning poses computation selection as a resource-rational process. But two challenges remain: (1) the optimal strategy is notoriously hard to solve, and (2) the underlying representations have to be hand-specified and symbolic.

Sixing Chen
@sixingchen.bsky.social

Our approach: we take a meta-RL algorithm recently proposed as a model of PFC, and train it to select not just physical actions, but also mental actions—internal computations that leave the environment unchanged but generate decision-relevant information.

Sixing Chen
@sixingchen.bsky.social

New preprint w/ @fredcallaway.bsky.social! How does the brain decide which computations to run? We combine rational meta-reasoning with a meta-learning algorithm to build a recurrent network that learns to select computations. www.biorxiv.org/content/10.6...

biorxiv.orgLearning to select computations in recurrent neural circuitsTwo hallmarks of biological computation are its flexibility and efficiency. These features are often attributed to cognitive control processes that balance external utility against computational cost. However, how the brain could implement such adaptive control remains unknown. Here, we provide one possible answer by combining the computational theory of rational meta-reasoning with a meta-learning algorithm recently proposed as a model of prefrontal cortex. This yields a recurrent neural network model that learns to select computations. In simple choice tasks, the model approximates the algorithms and representations of optimal symbolic models and reproduces neural dynamics observed in macaque orbitofrontal cortex. In multi-step planning tasks, the model replicates key behavioral signatures of human planning strategies and captures human neural dynamics associated with step-by-step mental simulation. Our framework unifies meta-reasoning and meta-learning by showing that learning to reason can be understood as learning to learn from information generated by one’s own cognitive operations, providing a mechanistic account of how adaptive control of thought can be implemented in neural systems. ### Competing Interest Statement The authors have declared no competing interest.
Sixing Chen
@sixingchen.bsky.social

Biological computation is remarkably flexible and efficient. This is often attributed to cognitive control over what to compute and when. But how is such control actually implemented?

Sixing Chen reposted

Fred Callaway
@fredcallaway.bsky.social

Despite the world being on fire, I can't help but be thrilled to announce that I'll be starting as an Assistant Professor in the Cognitive Science Program at Dartmouth in Fall '26. I'll be recruiting grad students this upcoming cycle—get in touch if you're interested!

You are up to date.