Neural Flip-Flops II: Short-Term Memory and Electroencephalography

Abstract

By enabling many brain structures state changes, the explicit cascaded oscillators proposed here can generate the rhythmic neural activity found in EEGs. The function of such synchronization in information processing systems is timing error avoidance. The narrow requirement for the oscillator input pulse duration suggests a possible relationship to the abnormal electrical activity characteristic of epileptic seizures. Together, flip-flops and synchronization by oscillators suggest a resolution to the longstanding controversy of whether short-term memory depends on neurons firing persistently or in brief, coordinated bursts. The proposed cascade of oscillators consists of a ring oscillator and four toggle flip-flops connected in sequence. The novel oscillator and toggle are composed of three and six neurons, respectively. Their operation depends only on minimal properties of excitatory and inhibitory inputs. The hypothesis that cascaded oscillators produce EEG phenomena implies that the distribution of EEG frequencies is determined by just two parameters, the mean (d) and standard deviation ({sigma}d) of the delay times of neurons that make up the initial oscillators in the cascades. For example, if d and {sigma}d are measured in milliseconds, the boundary separating the alpha and beta frequency bands is O_FD O_INLINEFIG[Formula]C_INLINEFIGC_FD With 4 and 1.5 ms being the best available estimates for d and {sigma}d, respectively, this predicted boundary value is 14.9 Hz, which is within the range of commonly cited estimates obtained empirically from EEGs. Four and 1.5 ms also accurately predict the peaks and other boundaries of the five major EEG frequency bands. Significance statementThe neuronal model proposed here implies several major aspects of electroencephalography. The matched periods of neural activity found in EEGs, and their wide distribution across the brain and across the frequency spectrum, follow from selective pressure for a biologically useful function: timing error avoidance for diverse brain functions in the trade-off between speed and accuracy. This activity can be achieved with a simple organization of synaptic connections and minimal neuron capabilities of excitation and inhibition. The multimodal distribution of EEG frequencies is an explicit function of the mean and variance of neuron delay times. The model suggests a relationship to epileptic seizures and a resolution to a short-term memory controversy. Two EEG characteristics make other models implausible.