Abstract:Perceptions and actions, thoughts and memories result from coordinated activity in hundreds or even thousands of neurons in the brain. It is an old dream of the physics community to provide a statistical mechanics description for these and other emergent phenomena of life. These aspirations appear in a new light because of developments in our ability to measure the electrical activity of the brain, sampling thousands of individual neurons simultaneously over hours or days. We review the progress that has been made in bringing theory and experiment together, focusing on maximum entropy methods and a phenomenological renormalization group. These approaches have uncovered new, quantitatively reproducible collective behaviors in networks of real neurons, and provide examples of rich parameter--free predictions that agree in detail with experiment.

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve This paper aims to explore how to describe collective behavior in real neural networks using statistical mechanics methods. Specifically: 1. **Combining Theory and Experiment**: - The paper reviews the progress made by combining theory and experiment, focusing on the maximum entropy method and the phenomenological renormalization group method. These methods reveal new, quantitatively reproducible collective behaviors in real neural networks and provide parameter-free predictions that are in detailed agreement with experiments. 2. **Application of New Experimental Methods**: - With the development of new experimental techniques, it is possible to simultaneously monitor the electrical activity of hundreds or even thousands of neurons. This explosive growth in data requires new analytical methods, thereby promoting interaction between theory and experiment. 3. **Maximum Entropy Method**: - The maximum entropy method provides a direct bridge from the new data of a large number of neurons to explicit statistical physical models. By measuring the average activity of individual neurons and pairwise correlations, an Ising spin glass model can be obtained to describe the activity patterns in the network. 4. **Renormalization Group Method**: - Inspired by the renormalization group ideas in condensed matter physics, researchers attempt to coarse-grain the activity patterns in large-scale neural networks. Preliminary results reveal precise scaling behavior, which in some cases can even be reproduced to the second decimal place. In summary, this paper attempts to understand the collective behavior in real neural networks using statistical mechanics methods and to establish precise links between theory and experiment through the maximum entropy method and the renormalization group method.

Statistical mechanics for networks of real neurons