Neural decoding

Neural decoding is a neuroscience-related field concerned with the reconstruction of sensory and other stimuli from information that has already been encoded and represented in the brain by networks of neurons. The main goal of studying neural decoding is to characterize how electrical activity of neurons elicit activity and responses in a dynamic world ^[1].

Encoding to decoding

Implicit about the decoding hypothesis is the assumption that neural spiking in the brain somehow represents stimuli in the external world. Neural decoding is completely useless if the neurons are firing randomly. The organism must perceive a set of salient stimuli that results in some internal learning. This learning is the object of neural decoding's desire. One must both understand how information is stored in the brain in the first place and how this information is used at a later point in time. This neural coding and decoding loop is a symbiotic relationship and the crux of the brain's learning algorithm. Furthermore, the processes that underlie neural decoding and encoding are very tightly coupled and may lead to varying levels of representative ability^[2][3]

Data Collection

Much of the neural decoding problem depends on the means by which data is collected. Previous methods relied on stimulating single neurons over a repeated series of tests in order to generalize this neuron's behavior ^[4]. New techniques such as high-density multi-electrode array recordings and multi-photon calcium imaging techniques now make it possible record from upwards of a few hundred neurons. Even with better recording techniques, the focus of these recordings must be on an area of the brain that is both manageable and qualitatively understood. Many studies look at spike train data gathered from the ganglion cells in the retina. Of all the possible subset of neurons to study, this particular area has the benefits of being strictly feedforward, retinotopic, and easily bottlenecked based upon the presentation of a stimulus to a particular subset of the visual field that the ganglion cells represent. The duration, intensity, and location of the stimulus can be controlled guaranteeing that a predetermined number of ganglion cells can be sampled within a significantly structured microcosm of the visual system.^[5] In addition to the visual system, other studies have used the discriminatory ability of rat facial whiskers as a medium for collected spike train data ^[6]

Probabilistic decoding schemas

When decoding neural data, arrival times of each spike t_1, t_2, ..., t_n = {t_i}, and the probability of seeing a certain stimulus, P[s(t)] may be the extent of the available data. The prior distribution P[s(t)] defines an ensemble of signals, and represents the likelihood of seeing a stimulus in the world based on previous experience. The spike times may also be drawn from a distribution P[{t_i}]. What we want to know is the probability distribution over a set of stimuli given a series of spike trains P[s(t)|{t_i}], which is called the response-conditional ensemble. What remains is the characterization of the neural code by translating stimuli into spikes, P[{t_i}|s(t)]; the traditional approach to calculating this probability distribution has been to fix the stimulus and examine the responses of the neuron. Combining everything using Bayes' Rule results in the simplified probabilistic characterization of neural decoding: P[s(t) | {t_i}] = P[s(t)|{t_i}] * (P[s(t)]/P[{t_i}]). An area of active research consists of finding better ways of representing and determining these probability distributions. ^[7]

Spike train number

The simplest coding strategy and also the lowest performing is the spike train number coding. This method assumes that the spike number is the most important quantification of spike train data. Given a set of stimuli, each stimulus should be represented by a unique firing rate across the sampled neurons. This strategy is purely spatial and assumes no correlation between one spike and another - each spike is pooled together into an overall count. (Equations needed)

Spike timing

Adding a small temporal component results in the spike timing coding strategy. Here, the main quantity measured is the number of spikes at a particular time t. This method adds another dimension to the previous. (Equations)

Temporal correlation

Temporal correlation code, as the name states, adds correlations between individual spikes. (Equations)

Agent-based decoding schemas

In addition to the probabilistic approach, agent-based models exist that capture the spatial dynamics of the neural system under scrutiny. One such model is hierarchical temporal memory, which is a machine learning framework that organizes visual perception problem into a hierarchy of interacting nodes (neurons). The connections between nodes on the same levels and a lower levels are termed synapses, and their interactions are subsequently learning. Synapse strengths modulate learning and are altered based on the temporal and spatial firing of nodes in response to input patterns. ^[8]

While it is possible to take the firing rates of these modeled neurons, and transform them into the probabilistic and mathematical frameworks described above, agent-based models provide the ability to observe the behavior of the entire population of modeled neurons. Researchers can circumvent the limitations implicit with lab-based recording techniques. Because this approach does rely on modeling biological systems, error arises in the assumptions made by the researcher and in the data used in parameter estimation.

Applicability

The advancement in our understanding of neural decoding benefits the development of brain-machine interfaces, prosthetics^[9] and the understanding of neurological disorders such as epilepsy.^[10]

References

1. ^ Jacobs, A. L., Fridman, G., Douglas, R. M., Alam, N. M., Latham, P. E., Prusky, G. T., & Nirenberg, S. (2009). Ruling out and ruling in neural codes Proceedings of the National Academy of Sciences of the United States of America, 106(14), 5936–5941. doi:10.1073/pnas.0900573106
2. ^ Chacron, M. J., Longtin, A., & Maler, L. (2004). To burst or not to burst Journal of computational neuroscience, 17(2), 127–136. doi:10.1023/B:JCNS.0000037677.58916.6b
3. ^ Boloori, A.-R., Jenks, R. A., Desbordes, G., & Stanley, G. B. (2010). Encoding and decoding cortical representations of tactile features in the vibrissa system The Journal of neuroscience : the official journal of the Society for Neuroscience, 30(30), 9990–10005. doi:10.1523/JNEUROSCI.0807-10.2010
4. ^ Plasticity of ocular dominance columns in monkey striate cortex. (1977). Plasticity of ocular dominance columns in monkey striate cortex.
5. ^ Warland, D. K., Reinagel, P., & Meister, M. (1997). Decoding visual information from a population of retinal ganglion cells Journal of neurophysiology, 78(5), 2336–2350.
6. ^ Arabzadeh, E., & Panzeri, S. (2006). Deciphering the spike train of a sensory neuron: counts and temporal patterns in the rat whisker pathway. The Journal of neuroscience.
7. ^ Rieke, F. (1999). Spikes: exploring the neural code. exploring the neural code (p. 395). The MIT Press.
8. ^ Hierarchical temporal memory: Concepts, theory and terminology. (2006). Hierarchical temporal memory: Concepts, theory and terminology. Whitepaper.
9. ^ Donoghue, J. P. (2002). Connecting cortex to machines: recent advances in brain interfaces Nature Neuroscience, 5 Suppl, 1085–1088. doi:10.1038/nn947
10. ^ Rolston, J. D., Desai, S. A., Laxpati, N. G., & Gross, R. E. (2011). Electrical stimulation for epilepsy: experimental approaches Neurosurgery clinics of North America, 22(4), 425–42, v. doi:10.1016/j.nec.2011.07.010

See also

• Correlation coding
• Independent-spike coding
• Phase-of-firing code
• Population coding
• Rate coding
• Sparse coding
• Temporal coding
• NeuroElectroDynamics