Time Is of the Essence: Neural Codes, Synchronies, Oscillations, Architectures

Abstract

Time is of the essence in how neural codes, synchronies, and oscillations might function in encoding, representation, transmission, integration, storage, and retrieval of information in brains. This Hypothesis and Theory article examines observed and possible relations between codes, synchronies, oscillations, and types of neural networks they require. Toward reverse-engineering informational functions in brains, prospective, alternative neural architectures incorporating principles from radio modulation and demodulation, active reverberant circuits, distributed content-addressable memory, signal-signal time-domain correlation and convolution operations, spike-correlation-based holography, and self-organizing, autoencoding anticipatory systems are outlined. Synchronies and oscillations are thought to subserve many possible functions: sensation, perception, action, cognition, motivation, affect, memory, attention, anticipation, and imagination. These include direct involvement in coding attributes of events and objects through phase-locking as well as characteristic patterns of spike latency and oscillatory response. They are thought to be involved in segmentation and binding, working memory, attention, gating and routing of signals, temporal reset mechanisms, inter-regional coordination, time discretization, time-warping transformations, and support for temporal wave-interference based operations. A high level, partial taxonomy of neural codes consists of channel, temporal pattern, and spike latency codes. The functional roles of synchronies and oscillations in candidate neural codes, including oscillatory phase-offset codes, are outlined. Various forms of multiplexing neural signals are considered: time-division, frequency-division, code-division, oscillatory-phase, synchronized channels, oscillatory hierarchies, polychronous ensembles. An expandable, annotative neural spike train framework for encoding low- and high-level attributes of events and objects is proposed. Coding schemes require appropriate neural architectures for their interpretation. Time-delay, oscillatory, wave-interference, synfire chain, polychronous, and neural timing networks are discussed. Some novel concepts for formulating an alternative, more time-centric theory of brain function are discussed. As in radio communication systems, brains can be regarded as networks of dynamic, adaptive transceivers that broadcast and selectively receive multiplexed temporally-patterned pulse signals. These signals enable complex signal interactions that select, reinforce, and bind common subpatterns and create emergent lower dimensional signals that propagate through spreading activation interference networks. If memory traces share the same kind of temporal pattern forms as do active neuronal representations, then distributed, holograph-like content-addressable memories are made possible via temporal pattern resonances.

Keywords: holographic memory; neural codes; neural networks; oscillations; radio communications; synchronies; temporal codes; timing nets.

FIGURE 1

Systems Overview. The general problem of understanding how nervous systems work as informational and experience-producing systems. (A) Different aspects of the problem [cf. Marr’s levels of analysis (Marr, 1982)]. (B) What is to be explained (explananda) in what terms (explanans). Reverse-engineering seeks to elucidate operant functional principles in nervous systems using information processing models that account for mental functions. Neuro phenomenal mappings predict subjective experiences by elucidating the neural correlates of states of consciousness (NCCs) and of the contents of consciousness (NCCCs).

FIGURE 2

Descriptions of time and temporal relations. (A) Linear, metrical time. (B) Cyclical time. (C) Ordinal time (temporal order of succession). (D) Wave time. (E) Future time (anticipation/prediction on the basis of past memories and present events). Symbols: Arrows time axes, E event timings, R reference timings, PE predicted events, A set of event-associated attributes.

FIGURE 3

Neural pulse code schemes. (A) A partial taxonomy of types of neural codes. At the triangle vertices are pure coding types, whereas the edges indicate combinations of two different types. P₁ and P₂ indicate two different spike patterns in the idealized spike raster icons. (B) Examples of other codes. Top left. A simple rate-channel “doorbell” code. Bottom left. Rate modulation code or coarse “temporal” code. Right. Burst length (# spikes per burst or duration of burst) and inter-burst interval (I₁ and I₂) codes. Rate modulation and burst-based codes don’t fit neatly into the taxonomy.

FIGURE 4

Temporal coding of pitch and timbre (vowel quality) in the auditory nerve. (A) Synthetic single formant vowel stimulus waveform (F0: 80 Hz, F1: 640 Hz, 60 dB SPL, 100 repetitions). (B) Post-stimulus time spike histograms of 42 auditory nerve fibers arranged by their characteristic frequencies (CFs). (C) Stimulus power spectrum. (D) Rate-place profile. (E) Stimulus autocorrelation function (ACF). (F) Population-interval distribution (PID) histogram of all-order interspike intervals of the whole ensemble. Delay intervals associated with major PID peaks closely correspond to the period of the perceived voice pitch (fundamental period 1/F0 = 12.5 ms). The pattern of minor peaks (0–5 ms) robustly encodes vowel formant structure and perceived aspects of timbre related to spectral shape. The PID provides a general purpose, neural representation of the stimulus autocorrelation function. The systematic first spike latency shifts in B are due to cochlear delays. From Cariani (1999).

FIGURE 5

Neural codes based on spike latencies relative to onset responses and oscillations. Onset responses of whole populations or subpopulations of onset responders can serve as a reference time. (A) Onset-based latency coding. First-spike latency, order of firing, or post-onset ensemble volley pattern can encode profiles of attributes such as relative intensity, by marking specific channels. (B) Time division multiplexing. Different time post-onset time windows encode different event attributes, here encoded by different response channels or combinations of them. (C) Phase-offset latency coding. Spike timings at different oscillatory phases encode presence of different attributes. Alternately successive oscillatory cycles can temporally discretize population rate responses to convert to rate-sequence codes. (D) Nested oscillations with common onsets (phase resets). Three common oscillatory frequencies are shown. Nested oscillations can serve as coding frameworks for onset-based latency codes, time-division multiplexing, and phase-offset coding. They can also selectively activate different subpopulations. They potentially provide a common temporal framework for hierarchical temporal grouping of events such as the recognition of spoken words, phrases, and sentences. Oscillation-based codes can convey multiple attribute distinctions.

FIGURE 6

Types of multiplexing of pulse-coded signals. (A) Frequency-division multiplexing in which different attributes (A1, A2) are represented by different spiking periodicities (interspike intervals). Intervals associated with the different attributes can either alternate or interleave within spike trains. (B) Code- or pattern-division multiplexing. Spiking patterns associated with different cutaneous and gustatory tongue sensations (Emmers, 1969, 1981). The proposed codes involve characteristic latency-interval patterns relative to onset bursts. Such codes could be multiplexed in single neurons, ensembles, and populations. (C) Synchrony-based multiplexing. Synchronized spikes can support binding of attribute combinations associated with separate objects. Here spike synchronies across channels bind together channels (C₁ – C₃) that code for specific attributes (A₁ – A₃) associated with different objects (O₁, O₂). (D) Oscillation-based multiplexing. Top waveforms depict two population oscillations (f1 = 4 Hz, F2 = 5 Hz). Summing the waveforms produces an additional beat periodicity at (f2 – f1 = 1 Hz). Multiplying them produces two beat periodicities or sidebands (f2 – f1 = 1 Hz; f1 + f2 = 9 Hz), characteristic of the cross correlation of the two oscillations. Multiplicative combinations of frequencies create new frequencies.

FIGURE 7

Proposed cortical latency-pattern coding framework for auditory events and speech reception. Onsets are neuronal bursts associated with transient acoustic contrasts. (A) Coding of perceptual attributes of auditory events by means of post-onset complex temporal spike patterns. (B) Coding of rhythmic patterns by temporal patternings of onset responses. (C) Coding of phones, syllables, words, and higher order syntactic and semantic relations in terms of onset-referenced latency-pattern codes in successive cortical regions. At each stage of the hierarchy, neural assemblies produce characteristic spike pattern markers associated with the recognition of acquired phonetic and linguistic distinctions.

FIGURE 8

Pattern-resonance scheme for neural signal dynamics. Complex temporal patterns of spikes percolate through synfire delay-coincidence networks activating pattern resonances, forming characteristic spike interference patterns. Grouping is achieved through common spike sub-patterns at local levels and event-rhythm patterns at global levels. Spreading activation creates provisionally stable standing wave patterns in recurrent synfire-chain delay loops. Temporal patterns associated with current goals are injected into the network to amplify task-relevant patterns and suppress irrelevant signals and circuits. Temporal patterns in reverberating short-term memory activate similar patterns and sub-patterns in long-term memory traces that encode “tape-recorder” event-and-reward sequences that serve as anticipatory guides for prospective behavior (Cariani, 2017). Schematic from (Cariani, 2015).

FIGURE 9

Schematic of a generic radio superheterodyne receiver.