We use physiologically-based neural circuit models and computational theory to study how the brain works. We are especially interested in understanding higher cognitive functions, with a focus on the prefrontal cortex. Our previous work has shown that a strongly recurrent microcircuit, with the NMDA receptor mediated local excitation that is balanced by synaptic inhibition, gives rise to persistent neural activity. Such self-sustained neural firing patterns underlie working memory, our ability to actively hold information in the mind in the absence of direct sensory stimulation. Furthermore, we found that a cortical local area capable of working memory, endowed with reward-dependent synaptic plasticity, can also subserve decision making computations and choice behavior, suggesting an unified `cognitive-type` cortical microcircuit organization. Other topics covered by research in my lab include the diversity of inhibitory cells in the cortex, synchronous oscillations, molecular switches in neurons and at single synapses. In particular, we and others have established that GABAergic inhibitory neurons play a fundamental role in the generation of a variety of coherent rhythmic activities in the thalamus and cortex. Currently, we are pursuing biophysically realistic large-scale circuit models of spiking neurons, in close collaboration with experimentalists, to elucidate general principles and detailed neurobiological mechanisms of key cognitive processes like working memory, decision making, selective attention, inhibitory control, as well as their impairments associated with mental disorders.