We are interested in brain mechanisms involved in high-level mental functions such as executive control, decision-making, and learning and memory (see Publications for individual studies). Our particular interests include mental functions that have developed through evolution and characterize us as humans. We thus use behavioral experiment and psychological assessment tools.
When we examine human brain functions, bioimaging technique must be non-invasive in most cases. We are mainly using functional MRI (fMRI) for a non-invasive neuroimaging technique. fMRI provides relatively high spatial resolution (a few mm), and more importantly, it allows us to measure neural-activity-related (BOLD) signals from the entire brain, even from deeper and smaller brain structures. Thus fMRI is still one of the most powerful imaging techniques to examine brain-wide functional mechanisms.
However, the BOLD signal involves a greater time constant (i.e. low temporal resolution; several secs), which is much slower than the transmission speed of action potential in neurons. This is because the BOLD signal reflects hemodynamic changes in blood capillaries around fired neurons. This physiological limitation sometimes becomes critical when we examine a series of mental and cognitive processes occurring in a brief period of time. In particular, examining dynamic aspect of brain functions including inter-regional interactions and directional signaling involved in such processes is one of the important challenges in fMRI studies.
In our prior studies, we examined temporal dynamics of fMRI signal during executive control (Jimura et al. 2010; Jimura & Braver 2010) and choice behavior (Jimura et al. 2013; Tanaka et al. 2020). These analyses were possible because the target time constants were greater than 10 secs. We are now aiming to develop analysis framework to shorten the time constant in fMRI analysis (Tsumura et al. 2021a; 2021b), which may allow us to examine causal mechanisms implemented in global brain networks involved in sophisticated mental functions.
On the other hand, the development of machine learning techniques has provided wide range of powerful tools for neuroimaging recently. We are particularly interested in the relationships between conventional standard fMRI analysis and those machine learning techniques that extract information about mental function s from MRI images (Jimura & Poldrack 2012; Jimura et al. 2014; Tsumura et al. 2021a). We are aiming to establish a novel framework of functional brain mapping by improving generalization performance of image classifier and using visualization of classification feature spaces for neuroimaging data (Tsumura et al. 2021b).
Our current projects include:
Value-based choice behavior.
Interaction of executive control and decision-making.
Time domain analysis of functional MRI signal.
Tanaka et al. J Neurosci 2020.
Tsumura et al. J Neurosci 2021.
Tsumura et al. Cereb Cortex 2021.