Although there is evidence suggesting that the inferior temporal (IT) cortex plays an important role in face processing and categorization, the exact neural mechanisms underlying these cognitive functions remain unknown. Here we address this issue by simultaneously recording the local field potentials (LFP) and single cells activity at 202 sites of the inferior temporal cortex (IT) of two macaques, while they passively fixated at pictures of monkey faces, human faces and objects. Our first goal was to investigate which features of the LFP, in frequency and time domains, were able to represent natural categories. For that, we calculated a selectivity index at two granularity levels: face vs. object (coarse selectivity) and monkey vs. human faces (fine selectivity). Our second goal was to study correlations between the selectivity of the LFP features and the selectivity of single cells recorded at the same sites. The data was first pre-processed as follows: for the LFPs we computed on each recording site: a) Visual-evoked-potentials (VEPs) and b) Single-trial-based instantaneous power and phase for different frequency bands. For the single cells we calculated i) Mean firing rate across trials and ii) Mutual information between stimulus classes and their associated responses (on each single-trial). Regarding selectivity of the VEPs, specifically the P100 deflection, we found that its onset latency occurred earlier for faces than for objects (p<0.01) and for monkey than for human faces (p<0.05). In contrast, the P100 amplitude did not systematically differentiate between these categories. In the frequency domain, we found that the degree of phase-locking (across trials in single electrodes) of the theta-band (4-8 Hz) around the P100 (80 ms to 120 ms after stimulus presentation) discriminated between faces/objects (p<0.01) and humans/monkeys (p<0.01). Considering correlations between selectivity of the LFP features and single cells, we found that ‘coarse’ (faces vs. objects) selectivity of the VEPs, particularly when using the amplitude of the N170 deflection, and also selectivity of the phase-locking of the gamma-band (low gamma: 28-48 Hz) around P100, significantly correlated (p<0.05) with the information about faces and objects of the single cells at those locations. More effects on correlations between LFP features