Neuronal Computation

Neuronal Computation:

Neuronal Computation: how neurons work together to produce behavior

Two interaction types:

• Spatial interactions = neurons interacting across space

  • Convergence

  • Lateral connections

• Temporal interactions = interactions over time

  • Delays

  • Feedback loops

Spatial = where neurons connect
Temporal = when/how activity changes over time

Convergence:

Convergence:

• Convergence = many receptors signals to one ganglion cell

• Signals summed together

• More receptors→ stronger response

• This increases sensitivity

With convergence

• Signals combine

• Ganglion firing increases as more receptors are activated

Without convergence

• Receptors stay separate

• One cell’s response does not increase just because more nearby receptors are active

Convergence = many inputs → one neuron → stronger summed signal

Receptive Fields:

Receptive Fields:

• Receptive field (RF) = the area of vision that a neuron responds to

• Usually roughly circular

• Stimulus inside the RF → neuron fires

• Stimulus outside the RF → little or no firing

• Spike train = the pattern of neuron firing

  • More spikes = stronger response

RF Size & Convergence

• More convergence → bigger receptive field

• Central vision → small RFs

  • Better detail

• Peripheral vision → bigger RFs

  • More convergence

  • Less detail

Center-Surround Organization

• Many RFs have a center and a surround

• The center and surround have opposite effects

• This comes from lateral interactions between neurons

Two common types:

• Excitatory center + inhibitory surround

• Inhibitory center + excitatory surround

Easy memory:
More convergence = bigger RF = less detail
Center and surround = opposites

Center-Surround Organization:

Center-Surround Organization:

• A receptive field often has a center and an area around it called the surround

• The center and surround have opposite effects on the neuron

ON-center / OFF-surround

• Light in the center → neuron fires more

• Light in the surround → neuron fires less

• Strongest response = bright center + dark surround

• This helps the brain notice contrast and edges

Easy memory:
Center ON = more firing
Surround OFF = less firing
Best response = bright center + dark surround

Lateral Inhibition:

Lateral Inhibition:

• Lateral inhibition = one neuron reduces the activity of nearby neurons

• In the retina, horizontal and amacrine cells help make center-surround receptive fields

Mach Bands:

Mach Bands:

• Edges can look lighter or darker than they really are

• This happens because there are different amounts of lateral inhibition near borders

Simultaneous Lightness Contrast

• The same gray color can look:

  • Darker on a light background

  • Lighter on a dark background

• The brightness around an object changes how bright it looks

Important Limitation

• White’s illusion shows that lateral inhibition does not explain every brightness illusion

Hermann Grid:

Hermann Grid:

• Hermann Grid illusion = gray/dark dots seem to appear at white intersections

• Dots seem to jump around in peripheral vision

Why it happens

• Caused by lateral inhibition

• Cells at intersections receive more inhibition

• More inhibition → less firing

• Less firing → intersection appears darker

Receptive Field Size

• Peripheral vision = larger receptive fields

  • More surround inhibition

  • Gray dots appear

• Central/foveal vision = smaller receptive fields

  • Less surround inhibition

  • Dots disappear when looked at directly

Intersection = more inhibition = darker dot
Peripheral = dots appear
Look directly = dots disappear

Mach Bands:

Mach Bands:

• Mach Bands = light and dark edges look more different than they really are

• Caused by lateral inhibition

• The visual system makes brightness differences at edges seem stronger

Near an edge:

• More inhibition → less neuron firing → looks darker

• Less inhibition → more neuron firing → looks lighter

• This creates extra light and dark bands near the border

• This is called edge enhancement

• It helps us see borders more clearly

Easy memory:
Mach Bands = lateral inhibition makes edges stand out more
More inhibition = darker
Less inhibition = lighter