Most neural experiments assume that each neuron is coding information independently. For example, if neuron A firing means red and neuron B firing means vertical then A and B together mean red and vertical. There is some evidence that in some systems, firing of neurons may be synergistic meaning that when two fire together (or with some fixed offset) there is a different meaning than that inferred from either alone. A first step in addressing this, is to find patterns of firing that occur more often than expected. Existing methods for pattern searching were limited in the type of pattern they could extract, assuming absolute reliability of spikes with temporal patterns of small numbers of neurons or considering only rate based codes. I extended the machine learning algorithm developed by Geoff Hinton and myself to look for stochastic patterns of neural firing that occur more often than expected. In awake but untrained owl monkey and marmoset monkey, we found temporal patterns of firing within and between neurons that occured more often than expected. Applying this algorithm in awake trained animals will allow us to investigate the correlation between these patterns and behavioral events - this is an exciting future direction of this research that we are eager to start in collaboration with Professor Chiba's lab.