小型專題研討會
Mini Symposia

Organizer: Dr. Yaron Caspi／國立臺灣大學心理學系
Abstract:
           Open Science is an approach to scientific research characterized by a set of research practices, hardware and software tools, open-access publication commitments, and specialized scientific mindsets that was introduced into science about 20 years ago. With each passing year, the Open Science community is becoming increasingly widespread and mainstream. The amount of funding resources for Open Science is constantly growing, efforts by international organizations to promote the approach are intensifying, educational courses associated with it are becoming more widespread, and Open Science infrastructure is expanding. This description is particularly applicable to Brain Sciences and Cognitive Neurosciences, which, due to the costs associated with conducting experiments and the potential replication crisis they face, stand at the forefront of the Open Science revolution. It is challenging to dispute the significance of this approach for Cognitive Neuroscience.

        Nevertheless, conducting research in accordance with the principles of Open Science is still more widespread in some countries and less common in others. In Taiwan, likely due to specific academic culture and ethical and bureaucratic barriers, Open Science practices are still not widespread. Since incorporating Open Science approaches into the Taiwanese cognitive neuroscience research agenda will benefit the Taiwanese community and help it better integrate into global research tendencies,  this relative lack of Open Science-oriented research calls for intensified efforts from national organizations associated with Neurosciences to put Open Science on their agenda.

        The current symposium aims to answer this call. Specifically, it seeks to introduce the Open Science approach to the Taiwanese Cognitive Neuroscience community and encourage more community members to engage with the field. This aim is achieved through a combination of meta-scientific related analysis, personal experiences, and a general introduction to the approach.

        The symposium is organized by a group of PIs and researchers working in Taiwanese Universities. This group was assembled a year ago (resulting from the efforts of Dr. Ju-Chi Yu), with the mission to expand and support the Open Science approach in Brain Sciences in Taiwan. Thus, the secondary aim of this symposium is to attract more PIs and researchers from the cognitive Neuroscience community to participate and collaborate with the efforts to make Taiwanese Cognitive Neuroscience as open as possible.

• Speaker 1: Assoc. Prof. Tzu-Yu Hsu（徐慈妤副教授）／國立中央大學生醫理工學院認知神經科學研究所 

• Topic 1: Introduction to Open Science and the Taiwan Open Brain Science Group

• Abstract:

As of today, numerous Open Science tools, programs, and services have been developed to support the growing Open Science community. Starting with a brief mention of one of the main drivers for Open Science, namely the replication crisis, the current talk will provide a concise overview of some key approaches, tools, and standardization mechanisms developed by the Open Science community. In particular, I will briefly introduce concepts such as pre-registration (e.g., through the Open Science Foundation), version-control programming (git), Brain Imaging Data Structure (BIDS), data sharing repositories (e.g., openNeuro), and collaboration format (BrainHacks).

In parallel, I will discuss the establishment of the Taiwan Brain Open Science Group last year, its aims, and its past, present, and future activities. Particularly, I will present the “Open Minds, Open Data workshop” we organized last summer, the inter-university student seminar for open data analysis we organized in December 2025, and the upcoming event we are organizing in the summer of 2026. Finally, I will present our primary communication tool for students and PIs through a specialized Discord channel, and call for increased participation of cognitive neuroscience community members in the activities of the Taiwan Open Brain Science Group.

• Speaker 2: Assoc. Prof. Niall Duncan／臺北醫學大學心智意識與腦科學研究所 

• Topic 2:  The Epistemic Value of Open Science Practices

• Abstract: 

The process of science requires that we make decisions based upon experimental evidence. These decisions may be pragmatic ones about direct action in the world, or they may be decisions related to theory development and falsification. In all cases, an element of the decision must be judging the epistemic value of the evidence. Evidence with low epistemic utility provides a poor foundation for any meaningful decision. As such, researchers should seek to maximize the epistemic utility of their work if they wish it to play a serious role in decision-making. One avenue for increasing such value is the adoption of open science practices. For example, preregistration of work provides the reader greater assurance that the evidence presented is not the product of post-hoc hypothesizing or p-hacking and so has more epistemic value. Similarly, the availability of the data and code upon which evidential tests are based provides transparency that can increase trust in any inferences made. In this talk, I will introduce these ideas and argue for the value of implementing them. As part of this, practical suggestions will be made to link in with the epistemic points raised. The target of the talk will be researchers at all stages of their career, but will particularly focus on early-career researchers.

• Speaker 3: Prof. Chun-Chia Kung（龔俊嘉教授）／國立中央大學認知神經科學研究所

• Title 3: How Do Open Science Practices Improve the Teaching of Neuroimaging Skills? What I’ve Learned from Analyzing fMRI Data from Openneuro.org as a Class Project

• Abstract: 

From 2020 onward, I have tried to analyze one openfMRI dataset in every fMRI class (NCKU course numbers and years: 2020 ds005 & ds1734, 2021 ds2770, 2022 ds0030, 2023 ds4562 & ds2837, 2024 ds5427, 2025 ds3604). Each time with 5-10 students, we managed to reproduce one or two paper figures (e.g., main results). In retrospect, what I have benefited from these practices are (a) the increased confidence in planning and carrying out data analysis in a reasonable time frame; (b) learning several new analysis methodologies, and (c) applying them to my own lab data to yield fruitful results. I will illustrate some recent examples (environmental concern, movie fMRI, and clinical/Ctrl classifications). Not only did I benefit from these practices, but students also benefit from doing real challenges and participating in exciting class projects. I will conclude with some painful realizations (e.g., don’t do 7T fMRI data analysis with current hardware!) and with the hope of leveraging the cross-campus exchange or collaboration in the future.

• Speaker 4: Dr. Yaron Caspi／國立臺灣大學心理學系

• Title 4: HARKing – Is It a Sin?

• Abstract: 

The move toward Open Science was driven, among other factors, by the replication crisis, the fact that much published research was non-replicable. It was suggested that the culprits for the replication crisis are a set of Questionable Research Practices (QRP) such as cherry-picking. HARKing (Hypothesizing After the Results are Known) is one of the identified QPRs. The practice refers to presenting the research hypothesis in the introduction or discussion sections as the original research hypothesis driving the research, even if this was not indeed the case at the time the experiment was conducted.

While there is a broad consensus that QRP, such as p-hacking, should be banned in Science, the discussion concerning HARKing is still ongoing, almost 30 years after the practice was first introduced. In this brief introduction to this QRP, I will discuss both arguments presented in favor of banning HARKing and arguments supporting HARKing. Supporters of HARKing claim that, under certain circumstances, it can actually promote Science instead of deterring it. In particular, I will emphasize arguments supporting HARKing from a Bayesian perspective and those related to philosophical discussions about the nature of Science.

Organizer: Prof. Shih-Chieh Lin（林士傑教授）／國立陽明交通大學神經科學研究所
Abstract:
           Understanding cognition requires explanations that link defined cell types and circuits to behavior. Rodent models provide that leverage, offering access to single-neuron dynamics and cell-type–specific manipulation at scale. This symposium assembles four researchers who dissect circuit mechanisms underlying decision making, attention, spatial navigation, and motor learning, using behaviors that span freely moving tasks to immersive virtual reality. Across talks, speakers will showcase large-scale population recordings (multi-region electrophysiology and imaging) combined with precise perturbations (e.g., opto/chemogenetics) to reveal how computations are distributed across identified circuits and cell classes. By extracting algorithmic principles from causal circuit tests, these studies deliver mechanistic accounts of cognitive functions at single-cell and population levels and offer a blueprint for cross-species convergence—linking circuit dynamics to cognitive theory and translational targets.

• Speaker 1: Prof. Shih-Chieh Lin（林士傑教授）／國立陽明交通大學神經科學研究所

• Topic 1: Selective Attention to Auditory and Visual Modalities Converges onto Noncholinergic Basal Forebrain Neurons

• Abstract:

Selective attention enables animals and humans to prioritize behaviorally relevant stimuli among competing sensory inputs. While the basal forebrain (BF) is known to modulate cortical activity and support attention, it remains unclear whether BF activity directly conveys an attention signal. Here, we show that selective attention to auditory and visual stimuli converges onto a shared population of noncholinergic BF neurons. Using a crossmodal task where rats rapidly switched attention between modalities, we found these neurons responded strongly to attended targets but weakly to the same stimuli when ignored, regardless of modality. These effects closely tracked both task-driven and spontaneous attention shifts on a single-trial basis. Moreover, BF responses reflected the linear summation of attended and ignored inputs, suggesting that sensory streams are filtered in parallel before converging in BF. These findings suggest that the BF may serve as a subcortical hub integrating attention signals across modalities to guide adaptive behavior.

• Speaker 2: Hao-Yun Teng（鄧皓芸）／中央研究院分子生物研究所 

• Topic 2:  A Low-Dimensional Code for Flexible Action in the Rodent Subthalamic Nucleus

• Abstract: 

The subthalamic nucleus (STN) is a central hub of the basal ganglia, yet how its neuronal populations encode motor behavior and internal state remains poorly defined. Using deep in vivo two-photon calcium imaging in behaving mice, we recorded single-cell activity in the STN during voluntary locomotion and reward-driven licking. Individual neurons exhibited mixed selectivity, responding to multiple behaviors with diverse temporal profiles. At the population level, however, STN activity collapsed onto a low-dimensional neural manifold with distinct axes encoding locomotor speed and licking intensity. Strikingly, a third population dimension captured a slow adaptation signal during sustained running, revealing an internal, time-dependent variable not predefined by overt behavior. Neural trajectories within this state space diverged depending on behavioral context, distinguishing self-initiated movement from reward-modulated transitions. Comparisons with the adjacent zona incerta demonstrated that this population geometry is specialized to the STN and supports more reliable decoding of motor variables. Together, these findings suggest that STN function is best understood as dynamic navigation through a low-dimensional motor–state space, providing a framework for interpreting how deep brain stimulation may restore movement by steering population trajectories rather than modulating single signals.

• Speaker 3: Assoc. Prof. Wen-Kai You（游文愷副教授）／國防醫學大學腦科學研究中心

• Title 3: Linking Mouse and Primate Attention Networks: Circuit Architecture and Visual Response Analysis of Cg/M2 in Mice

• Abstract: 

The frontal eye field (FEF) of primates plays a critical role in controlling attention and eye movements, yet how these functions are implemented in neural circuits remain poorly understood. Investigating FEF functions in a mouse model could provide new insights, as mice offer a wide range of genetic and viral tools, enabling precise labeling and manipulation of neural circuits not feasible in primates.

Despite this potential, the mouse brain region corresponding to the primate FEF has not been conclusively identified. We hypothesize that the cingulate cortex (Cg) and adjacent secondary motor cortex (M2) in mice serve as functional analogues of the primate FEF because: (1) Cg/M2 sends long-range projections to the visual cortices, and (2) modulates the tuning properties of visual neurons. (3) Stimulating Cg/M2 evokes saccadic eye movements, and (4) enhances visual discrimination performance of mice. All these features parallel those of the primate FEF.

To further investigate how Cg/M2 modulates visual neurons and improves visual performance, we employed viral tracing and tissue-clearing techniques to examine connectivity patterns between Cg/M2 and two key nodes of the attention network: the primary visual cortex (V1) and the superior colliculus (SC). Additionally, to assess whether Cg/M2 neurons exhibit visual response properties similar to those of FEF neurons, we conducted electrophysiological recordings in awake mice to characterize the responses of Cg/M2 neurons to a series of visual stimuli. These findings will help establish a clearer cross-species framework for studying attention and eye movement control, facilitating the use of genetic and viral tools in mice to probe the neural mechanisms underlying FEF functions.

• Speaker 4: Dr. Ching-Lung Hsu（徐經倫博士）／中央研究院生物醫學科學研究所

• Title 4: Large-Scale Single-Neuron Dynamics Reveals Cognitive Regularities in Cortico-Hippocampal Transformations

• Abstract: 

A powerful brain theory approach for how autobiographical memories are encoded is to frame the neural activity as a conjunctive representation of space and other non-spatial context. Hippocampal Cognitive Map states that the hippocampus does so to enable episodic memory, which supports flexible planning. At its core, it requires a functional mechanism of hippocampal pyramidal neurons to integrate spatial and non-spatial inputs, transforming them into a flexible spatial code reflecting the sequences of prior experience according to space and time (the “algorithm”).

What is the core neural mechanism underlying such an algorithm? We argue that recording made only from the hippocampus per se cannot solve the implementational problem. Essentially, it is a question that asks for specifying the inputs and outputs to delineate the computations of hippocampal place cells during goal-directed memory tasks. In this talk, I will first describe the details of the hippocampal algorithm, in the form of “state machine”, as hinted by latest comparisons made between theoretical models and large-scale physiological recordings. In our own work, anatomically and functionally connected single neurons of the hippocampus, entorhinal cortex and sensory cortices were simultaneously recorded, using high-density electrodes Neuropixels 2.0, from mice performing two-choice tasks in well controlled virtual-reality (VR) environments. We will discuss to what extent hippocampal states which dynamically “split” according to behavioral experiences is genuinely generated only in the hippocampus, based on large-scale single-neuron evidence. I believe this effort may offer a neural justification for functional algorithms that implement a hippocampal state machine as well as a cognitive map.