In this talk I will present some results from recent investigations in phonetic neuroscience. The aim of the research is to discover the neural code of phonetics. The data come from recordings from a dense grid of electrodes placed directly on the surface of the brain in patients who were undergoing surgery for epilepsy. These neural imaging data have fine resolution in both time and frequency and have low enough noise that a relatively few trials is needed in order to find rich phonetic information. In the first of the three studies I will describe, we found that during speaking the motor cortex shows patterns of activity that group sounds by articulator – labials are similar to other labials, dorsals are similar to other dorsals, etc (Bouchard et al., 2013, Nature 495, 327-332). In the second study, we found that during listening the superior temporal gyrus shows patterns of activity that group sounds by manner of articulation – stops are similar to stops, fricatives to fricatives, etc. (Mesgarani et al., 2014, Science 1006-1010). The third study examines a pattern of activity in the motor areas of the cortex that appears during listening (Cheung et al., 2016, eLife 2016;5,e12577). This pattern has been noted by prior researchers who have speculated that motor activity during speech perception suggests the activity of mirror neurons and perhaps that the motor theory of speech perception is supported. Our data complicate this interpretation because we are able to decode the phonetic information in the motor area during perception and what we find is surprising. The pattern of activity is like the pattern found by Mesgarani et al. in STG – sounds are grouped with each other by manner of articulation, not by articulator. I’ll discuss the implications of this particular finding, and the broader implications of neuroscience for linguistics.