The phenomenon of dual coding

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

It was once thought that neurons use firing rate along to communicate. Now a number of possible neural codes have been proposed. Discuss evidence that suggests different populations of cells use different neural codes, and that some may even be ‘bi-lingual’.


The study of neural coding has been a very important research topic in the neuroscience field. Although the neuronal coding is not fully understood yet, but it has been known for a long time that the information is conveyed between neurons by the sequence of action potentials or spikes. The action potential is a brief and uniform pulse of electrical activity. These spikes travel along the axon to reach several postsynaptic neurons creating the postsynaptic potentials. The postsynaptic neuron may receive many spikes from surrounding neurons within a short time. In this case the neuron membrane potential will reach a critical value to trigger a spike which will be transmitted to other neurons.

Different neural coding has been suggested in which information are encoded in neural signals, the most important of them being rate coding, temporal coding, population coding, sparse coding and correlation coding. Other neural coding have been suggested.

Rate coding refers to information being carried by the firing rate. Firing rate is the number of spikes in a given time window. In temporal coding the exact time at which an individual spike occurs is the most important event. Population coding refers to information available from ensembles that go beyond simple summation of individual signals. Sparse coding refers to strong selectivity so that relatively small numbers of neurons would be involved in carrying the signal. In correlation code one spike can alter the meaning of another. Recent studies suggest the possibility of dual or multiple codes, transformation between the neural codes, simultaneous use of different codes and using different codes by different population of neural cells.

In the present essay a short view of the mentioned topics are presented.

Rate code

The rate code was initially shown by(Adrian and Zotterman 1926) when the number of spikes emitted by the receptor neuron increased with the force applied to the muscle. Another example is the touch receptor in the leech, the stronger the touch stimulus, the more spikes occur during a stimulation period of 500 ms (Kandel and Schwartz, 1991).

In a study of CA1 pyramidal cells, a more monotonic increase in firing rate was observed with lowest firing rate at the beginning of the place field and highest rate at the end ( MEHTA MR, QUIRK MC, AND WILSON MA Neuron 25: 707–715, 2000).

Temporal code

In temporal coding, a neuron encodes information through the precise timing of spikes on a millisecond time scale.

Mainen and Sejnowski (Science 1995, 268, 1503-1506 ) in an examination of the reliability of spike generation using recording from neurons in rat neocortical slices found that the precision of spike timing was depended on stimulus transients. Stimuli with fluctuations resembling synaptic activity produced spike trains with timing reproducible to less than 1 millisecond while constant stimuli led to

imprecise spike trains.

The Temporal Coding of Visual Information in the Thalamus was shown by Pamela Reinagel and R. Clay. They recorded from individual neurons in the cat lateral geniculate nucleus (LGN) while presenting randomly modulated visual stimuli. The responses to repeated stimuli were reproducible, whereas the responses evoked by non-repeated stimuli drawn from the same ensemble were variable. Stimulus-dependent information was quantified directly from the difference in entropy of these neural responses. Information rate was correlated with the firing rate of the cell, for a consistent rate of 3.6 6 0.6 bits/spike (mean 6 SD). This information was attributed to the high temporal precision with which firing probability was modulated; many individual spikes were timed with better than 1 msec precision.

Population coding

A population code is defined where neurons jointly represent information, therefore statistical dependence is expected. A frequently touted advantage of population coding is that it suppresses the effects of neuronal variability (L. F. Abbot, Peter Dayan, Neural Computation 11, 91–101, 1999 ).

A. P. GEORGOPOULOS et al (Science 233, 1416 (1986) showed that although individual neurons in the arm area of the primate motor cortex were only broadly tuned to a particular direction in three-dimensional space, the animal can very precisely control the movement of its arm. The direction of movement was found to be uniquely predicted by the action of a population of motor cortical neurons. When individual cells were represented as vectors that make weighted contributions along the axis of their preferred direction, the resulting vector sum of all cell vectors (population vector) was in a direction congruent with the direction of movement.

Einat Granot-Atedgi et al(March 14, 2013, OPLS )have shown that in a population of 100 retinal ganglion cells in the salamander retina responding to temporal white-noise stimuli, dependencies between cells play an important encoding role.

The encoding of sensory information by populations of cortical neurons forms the basis for perception (R.A.A.I., S.P., and C.K., The Journal of Neuroscience 2013, 33(46): 18277-18287). It was found that when choosing subpopulations randomly from the recorded ensemble, the average population information increases steadily with population size. This scaling was explained by a model assuming that each neuron carried equal amounts of information, and that any overlap between the information carried by each neuron arises purely from random sampling within the stimulus space.

Sparse coding

The sparse coding of a signal is one in which only a small fraction of the neurons in a network is activated by a given stimulus (Edmund T Rolls and Alessandro Treves, Network l(1990) 407-421)

William E. Vinje and Jack L. Gallant (SCIENCE VOL 287 18 FEBRUARY 2000) studied the sparse code of primary visual cortex (area V1) which was investigated by recording from V1 neurons in awake behaving macaques during both free viewing of natural scenes and conditions simulating natural vision. Stimulation of the nonclassical receptive field increases the selectivity and sparseness of individual V1 neurons, increases the sparseness of the population response distribution, and strongly decorrelates the responses of neuron pairs. These effects were attributed to both excitatory and suppressive modulation of the classical receptive field by the nonclassical receptive field and did not depend critically on the spatiotemporal structure of the stimuli. During natural vision, the

classical and nonclassical receptive fields function together to form a sparse

representation of the visual world.

Correlation code

A code in which one spike can alter the meaning of another is called

a "correlation code" to indicate that correlations between different spikes carry information. These spikes may be fired by one neuron or by different neurons. (Neural Codes and Distributed Representations, Foundations of Neural Computation, L. Abbott and T. Sejnowski MIT Press (1999))

Adam L. Jacobsa et al (5936–5941 _ PNAS _ April 7, 2009 _ vol. 106 _ no. 1, set up a strategy to determine which codes are viable, and applied it to the retina as a model system. They recorded from all the retinal output cells an animal uses to solve a task, evaluated the cells’ spike trains for as long as the animal evaluates them, and used optimal, i.e., Bayesian, decoding. The results showed that 2 of the codes, the spike count and spike timing codes, did, in fact, fall short. The performance of the spike count code fell substantially short. This result also held when spikes were counted in windows smaller than the length of the stimulus presentation, indicating that the failure of this code was not being exaggerated by counting spikes in the full 300-ms window. Even when spikes were counted only in 100 ms and 50 windows, the spike count code performed substantially worse than the animal. The second result was that the spike timing code also fell short, though, that the failure of this code was much less than that of the spike count code. The last result was that the temporal correlation code did perform as well as the animal.

The Theta-Gamma Neural Code

The cross coupling of theta and gamma waves form a code for representing multiple items in an ordered way. This form of coding has been demonstrated in the hippocampus. Different spatial information in the hippocampus is represented in different gamma sub cycles of a theta cycle. It has been suggested that this coding scheme coordinates communication between brain regions and is involved in sensory as well as memory processes (John E. Lisman,* and Ole Jensen,* Neuron 77, 2013; John Lisman and Gyo¨rgy Buzsa´ki, Schizophrenia Bulletin 34, 974–980, 2008 ).

Transient coding

Transient coding considers that transactions among neuronal systems use transient dynamics that are distributed in a structured way over both space and time, suggested by Karl J. Friston (NEUROIMAGE 5, 213–220 (1997) ).

In contrast to synchronization models, transient coding does not depend on interactions at the same frequencies, in different parts of the brain, but involves covariations among different frequencies and can therefore be considered a more general form of coding.

The phenomenon of dual coding

The idea of related processing mechanisms in the brain for spatial and temporal informa­tion has been suggested recently ([1] E.M. Hubbard, et al., Nature Neurosci. Rev. 6 (2005) 435–448. [2] V. Walsh.Trends Cogn. Sci. 7 (2003) 483–488).

Recent experimental and theoretical work showed that population rate code and a synchronous code, can be dually used in a single architecture (Masuda and Aihara 2007). They found that depending on the types of feedback coupling and shared connectivity, clusters are independently engaged in population rate coding or synchronous coding, or they interact to serve as input filters.

Hengtong Wang et al (Journal of Theoretical Biology 328(2013)19–25) have studied the First Spike Latency (FSL) in 3 different classes of neurons. With changes in the input parameters, the FSLs of the class1and 2 neurons exhibit similar properties, but the FSL class3 neurons became slightly longer and only produce responses for an arrow range of initial phase if input frequencies are low. They showed that FSL and firing rate responses are mutually independent processes and that neurons can encode an external stimulus into different FSLs and firing rates simultaneously, consistent with the current idea of dual or multiple coding mechanisms.

Bush D et al [ (2010) PLoS Comput Biol 6(7)] investigated the phenomenon of dual coding in the hippocampus by examining a spiking recurrent network model with theta coded neural dynamics and an STDP rule that mediates rate-coded Hebbian learning when pre- and post-synaptic firing is stochastic. They have demonstrated that this plasticity rule can generate both symmetric and asymmetric connections between neurons that fire at concurrent or successive theta phase, respectively, and subsequently produce both pattern completion and sequence prediction from partial cues.

Christopher Burgess et al (Space, Time and Number in the Brain. 2011, 59-69) have reviewed the potential for neuronal oscillations to produce phase codes for information that can be derived from simple sensory/motor inputs by a process of tempo­ral integration. Thus it is possible that distance traveled through a place cell’s firing field is encoded by its phase of firing relative to the ongoing theta rhythm, given an input that codes for running speed. Equally the spatial pattern of firing of grid cells may reflect temporal integration of inputs reflecting running velocity along specific directions.

VICTORIA BOOTH AND AMITABHA BOSE (J Neurophysiol 85: 2432–2445, 2001) has discussed the neural mechanism in Model CA3 pyramidal cell for generation of rate and temporal codes. They have shown that, depending on its timing, a short dose of fast decaying synaptic inhibition can either delay or advance the timing of firing of subsequent bursts. Moreover, increasing the strength of the inhibitory input is shown to modulate the burst profile from a full complex burst, to a burst with multiple spikes, to single spikes. It is also shown how slowly decaying inhibitory input can be used to synchronize a network of pyramidal cells.

Simultaneous use of dual and multiple codes:

A single neuron or a population of neurons are able to use more than one code simultaneously.

The evidence for a bias in neuronal circuits toward temporal coding and the coexistence of rate and temporal coding during population rhythm generation is presented by Matt Ainsworth et al (572 Neuron 75, August 23, 2012) in which the coincident expression of multiple types of gamma rhythm in sensory cortex suggests a mechanistic substrate for combining rate and temporal codes on the basis of stimulus strength.

P. Knüsel et al ( Network: Computation in Neural Systems, 2007, 18: 1, 35 — 62 ) have used optical imaging of the projection neurons of the moth antennal lobe and were able to establish quantitatively that spatio-temporal encoding strategies is used by the insect brain.

Studies on mammalian hippocampus (Ahmed and Mehta 2009) have revealed that CA1 pyramidal cells are activated when the animal is in at a selective spatial location by producing spike trains in a “rate code” paradigm. But when the rat runs they recorded an 8 Hz theta oscillation from the CA1 region as well. The first spike occur late in a theta cycle and as time passes the spike appear earlier in each cycle until it appears almost to the beginning of the cycle. The theta phase is a “temporal code” since the information about the position of the rat is encoded in the spike timing and not the numbers of spikes.

Transformation of temporal and rate coding

E. Ahissar et al (NATURE | VOL 406, 2000) has shown transformation of temporal to rate coding in a somatosensory thalamocortical pathway. In response to varying stimulus frequencies, the lemniscal neurons exhibited amplitude modulations and constant latencies. In contrast, paralemniscal neurons in both thalamus and cortex coded the input frequency as changes in latency. Because the onset latencies increased and the offset latencies remained constant, the latency increments were translated into a rate code: increasing onset latencies led to lower spike counts. Variable latencies and effective cortical feedback in the paralemniscal system can serve the processing of temporal sensory cues, such as those that encode object location during whisking.

X. WANG et al (Neuroscience 154 (2008) 294–303) was able to show that the primary auditory cortex (A1) uses a temporal representation to encode slowly varying acoustic signals and a firing rate– based representation to encode rapidly changing acoustic signals, also the dual temporal-rate representations in A1 represent a progressive transformation from the auditory thalamus, and firing rate based representations in the form of monotonic rate-code encode slow temporal repetitions in the range of acoustic flutter in A1 and more prevalently in the cortical fields rostral to A1 in the core region of marmoset auditory cortex, suggesting further temporal-to-rate transformations in higher cortical areas.

T. Taillefumier and M. O. Magnasco (PNAS 2014, doi/10.1073/pnas.1212479110)have studied the neural coding transition theoretically. They have found that a phase transition in the first passage of a Brownian process through a fluctuating boundary, the critical point Hc = 1/2 corresponds to a widely studied case in the theory of neural coding, in which the external input integrated by a model neuron is a white-noise process, as in the case of uncorrelated but precisely balanced excitatory and inhibitory inputs. They argue that this transition corresponds to a sharp boundary between rate codes, in which the neural firing probability varies smoothly, and temporal codes, in which the neuron fires at sharply, defined times regardless of the intensity of internal noise.

Different cell population use different neural code

X. WANG et al (Neuroscience 154 (2008) 294–303) in the study of neural coding of temporal information in auditory thalamus and cortex has found that A1 neurons are responsive to both rapid changes within a short time window and the inter-stimulus intervals suggesting that discharge rate–based mechanisms are responsible for encoding rapid time varying signals. The two populations of A1 neurons, referred to as synchronized and non-synchronized populations, appeared to encode repetitive stimuli by spike timing and average discharge rate, respectively. Neurons in the synchronized population showed stimulus-synchronized discharges at long inter-stimulus intervals, but few responses at short ICIs. This population of neurons can thus represent slowly occurring temporal events explicitly using a temporal code. The non-synchronized population of neurons did not exhibit stimulus-synchronized discharges at either long or short ICIs. This population of neurons can implicitly represent rapidly changing temporal intervals by their average discharge rates.

Stefan Leutgeb1 and Jill K. Leutgeb (14:745–757 © 2007 Cold Spring Harbor Laboratory Press 745 Learning & Memory) have shown that

CA3 cells are bound to a large degree to a spatial coordinate system, while

CA1 cells can become more independent of a map-based mechanism and allow for a larger degree of arbitrary associations, in the temporal domain. The CA3 network can rapidly alter its firing rate in response to novel sensory inputs and is thus not as strictly tied to spatial mapping as grid cells in the medial entorhinal cortex.

Marsat G and Maler L. (J Neurophysiol 104: 2543–2555, 2010) have shown that communication signals used during courtship (big chirps) and during aggressive encounters (small chirps) are encoded by different populations of ELL pyramidal cells, namely I-cells and E-cells, respectively. It is shown that the encoding strategy differs for the two signals. These differences allow these cell types to encode specifically information rich features of the signals. Small chirps are detected, and their timing is accurately signaled through stereotyped spike bursts, whereas the shape of big chirps is accurately represented by variable increases in firing rate.