Datensatz

Zusammenfassung

The coding properties of sensory neurons are typically examined after aligning neural and sensory events relative to a laboratory-based temporal reference, such as the computer clock controlling sensory stimulus presentation and data acquisition. While this approach often reveals considerable stimulus information in finely timed response patterns (e.g. in millisecond precise spike timing), it remains unclear whether the brain may exploit such temporal information as it does not have access to the laboratory-based temporal reference frame. Rather, the brain has to rely on an intrinsic temporal reference frame to decode temporal patterns of spiking responses. This raises the questions of whether, and how, the brain can compute temporal patterns of neural responses based on an internal time reference frame. One candidate mechanism for such an internal reference frame is provided by slow rhythms, which are ubiquitous in sensory cortical local field potentials (LFPs). We here show that theta-band rhythms in sensory cortices have the key features to serve as an internal reference for the efficient readout of informative temporal response patterns. To this end we analyzed neural responses to naturalistic stimuli recorded in auditory and visual cortices of macaque monkeys using computational methods of stimulus decoding and information theory. We found that the phase of theta LFPs (4-10Hz) can be used to group action potentials of single neurons and populations thereof into response patterns that carry similar levels of single trial stimulus information as obtained using the laboratory-based reference frame: using a phase-based reference frame recovered 70-90 of the information available when decoding the same responses using the laboratory based reference frame. This is possible because slow rhythms robustly entrain to the sensory environment both during isolated stimuli and during continuous stimulation periods. Importantly, this phase-based reference frame allows for sufficient temporal precision to create phase-aligned responses on those time scales on which neural responses are most informative about complex stimuli (here a few tens of milliseconds). These findings provide a direct computational proof of concept for the hypothesis that slow rhythmic activity may serves as internal reference frame for information coding and thereby bolster speculations about a mechanistic function of slow rhythmic activity in sensory cortical information coding.