Neural Coding

Question 1

Basics (what is neural coding)

Answer 1

—> neural encoding refers to the map from stimulus to response

—> aims to understand how neurons respond to a wide variety of stimuli and to construct models that can predict responses

—> there are several hypothesised coding schemes

Question 2

Place coding (overview)

Answer 2

= the location of the neuron encodes information about stimulus

• neuron 1 responds to sound 1 etc.
• can be represented in various topographic maps

Tonotopic map is passed from the basilar membrane to the auditory nerve fibers via hair cells

tonotopy = spatial arrangement of where sounds of different frequency are processed in the brain. Tones close together are represented topologically neighboring regions in the brain

• the cochlear vibrates in response to sound along its length
• narrow and more frigid at beginning, becomes increasingly wider and more flexible

—> because the incoming sound wave is broken down into different frequencies

• basilar membrane within cochlea contains hair cells (inner and outer)

–> outer hair cells: amplification of sounds to increase sensitivity

–> inner hair cells: respond to the amplified vibrations and excited neurons to which they are connect

–> hair cells move in response to the wave being transported through the cochlea

Question 3

Best frequency (also complex sounds)

Answer 3

—> each neuron has a best frequency where they fire

• together they cover the hearing range of an animal
• for low frequencies, neurons are tuned to a wider range of frequency

—> leads to poor discrimination in these frequencies

• for high frequencies, neurons are more fine tuned in order to identify that the sound has changed

Complex sounds

–> most have more than one frequency

—> basilar membrane essentially performs a Fourier Transform in order to decompose the sound waves into its frequency components

—> this means that several neurons will be active at the same time, firing together in a pattern

—> decomposed frequencies can be represented in a sound spectrogram which shows fluctuations of frequencies and energies over time

—> tonotopic order of sound frequency is maintained in each brain area of the auditory system (but the further down the path, the less neurons will respond to a pure tone)

Question 4

Rate Coding

Answer 4

• value of a stimulus is represented by how many spikes a neuron fires
  → as intensity increases frequency of AP increases
• neurons can respond to a wider range of frequencies if the intensity is high but still has an ideal frequency
• the higher intensity of the stimulus, the more spikes
• same as in visual system where greater firing rate is associated with greater contrast of stimuli
• rate coding assumes that most information about stimulus is contained in the firing rate of the neuron
• any information that could be encoded in the temporal structure of spikes is neglected

—> rate coding may be too simplistic to describe brain activity

BUT this leads to confound between frequency and sound level code

—> instead better to look at population code

original experiment: they had a muscle and they hung weights from the muscle and recorded neural response

—> the more weights they were hanging on the muscle, the greater the spikes

Question 5

Population Code

Answer 5

• represents stimulus value not by the output of a single neuron but instead through the pattern of responses across a population of neurons

—> represents stimuli by using the activity of several neurons

• response of many neurons combined to determine information about the input

—> single neuron has a distribution of responses

• looking at the overall population response shows a more complete picture of the frequency and intensity of the sound

—> as single neurons may fluctuate in their response

• e.g. visual medial temporal area, neurons encode the direction in which an object is moving
• advantages of this type of coding:

–> reduction of uncertainty due to neuronal variability

–> ability to represent a number of different attributes simultaneously

–> much faster than rate coding, can reflect changes almost instantly

Question 6

Temporal Code

Answer 6

= frequency of spikes

• a neural code is identified as a temporal code when precise spike timing and high-frequency firing rate fluctuations are found to carry information
• when neurons exhibit high-frequency fluctuations of firing-rates, rate coding models suggest that these are noise while temporal coding models suggest that they encode information
• if the brain only used rate codes to convey information, it would make evolutionary sense that firing rates were consistent.

—> instead temporal code explains the “noise” as something that encodes information and affects neural processing

Two types of code

• first spike latency —> different stimuli lead to spikes at different times —> stimuli can be differentiated by time
• interspike interval code —> different stimuli lead to spikes in different intervals → stimuli can be differentiated by length in intervals

Example

• temporally encoded information can help to discriminate between two different tastes of the same category (sweet, bitter..) that elicit very similar spike counts

—> temporal pattern may be used to determine its identity

• but at the same time rate coding is used as well in order to differentiate between basic tastant types —> rate and temporal codes contain different information

Question 7

Phase Locking

Answer 7

—> frequencies can be encoded through the timing of spikes called phase locking
→ combines spike count code with a time reference based on oscillations
—> important because neurons in some cortical sensory areas encode rich stimuli in terms of their spike times relative to phase of ongoing oscillatory fluctuations rather than only in terms of their spike count
- sine form of sound wave lines up with movements of hair cells and consequently with spikes in auditory nerve
- this means that timing of spikes becomes locked to the phase of the sound
—> provides temporal code of sound’s frequency