W4 Receptive Fields and Convolution

Whats the receptive field of a retinal ganglion cell like?

• Bipolar cells and interneurones change the signal/output they receive from photoreceptors to create an excitatory or inhibitory response to light

• If we shine a light into the center we get an excitatory response and cell starts to fire- if we shine a light into the inhibitory centre the firing rate decreases

Whats spontaneous discharge?

50 pps- base firing rate of cells at all times

What happens when light shines on the excitatory centre (centre of retinal ganglion cell?)

• Add +50pps onto baseline

What happens if light is shone off the edge of the excitatory centre?

Still prompts a response but not as high- 25pps

What happens when light is shone in the inhibitory surrounding?

Response of the retinal ganglion cell falls below spontaneous discharge- -20pps, which leads to 30pps action potential

How are graded potentials within the retina formed?

Light is transduced (turned from) electromagnetic energy a chemical response from neurotransmitters to get neurotransmitters

Why do we generate action potentials to the brain from RGC’s?

Because they’re better for going long distances to the brain

What do different parts of a spike train represent?

• Transient response when light is shone in the center of the RGC

• Sustained response and eventually spontaneous discharge when stimulus is turned off

How do you read a plot cell response?

• Spontaneous discharge is baseline rate

• When we turn the stimulus on rate increases, increasing number of action potentials being generated

• When light is in excitatory place rate falls to sustained response

What does a plot cell response look like in the inhibitory surround?

• No transient response- simply dropped sustained response

What does the plot cell response look like when we shine light on the whole cell?

• Start with spontaneous discharge

• Small stimulus onset- returns back to baseline

• When we turn stimulus off we get a small drop below baseline rate- quickly returns

• Transient response is VERY small

Whats the response of a light spot in an excitatory region?

Higher firing rate

Whats the response of a dark spot in an excitatory region?

Reduced firing rate

Whats the response of a light spot in an inhibitory region?

Firing rate falls

Whats the response of a dark stimulus in an inhibitory region?

Increase firing rate

What is convulution?

• Across the receptive fields we’re getting various signals which tells us ab the presence of an edge

• Convolution involves taking the retinal image (various amount of stimulus from the photoreceptors)

• These feed into the RGC’s with their various inhibitory/excitatory responses

• Receptive fields of RGC’s give us the pattern which produces a neural image

What is an on-centre cell?

Excitatory centre, inhibitory surround

How does the Hermann grid work?

• The on centre cell is completely illuminated in the centre

• The inhibitory centre cells receptive field is getting more input than the edges between the intersections

• The strength of the signal coming from the RGC at the intersection is gonna be lower as we have more light in the inhibitory part

Why do we see black dots in the Hermann grid?

• At the intersections we have a relatively lower signal so it must be darker

Why do we not get dots at the intersection we’re looking at?

• receptive fields of the foveal cells are very small so the illusion is destroyed

What number does the absence of light represent?

• Black is 0

• White is any big number (1000)

• Grey is anything in between

How do we calculate a single grey level?

Multiple input image by weighing function (luminance x Gaussian function) and calculate the sum

How do we represent different spacial scales?

• Narrow is a fine scale with detail, broad is a coarse scale which is good for overall impression

What is filtering?

• To convolve the input image with the weighing function to generate the output image

• E.g- fine detail is progressively filtered out as we increase the spread of the Gaussian weighing function

Whats the difference between convolution and filtering?

Convolution is just a way of filtering- essentially multiplying then adding up values

What do RGC’s do?

• Operate at various spacial scales, highlight edges and remove redundancy

• Edges are found at the transition between light and dark

How do we know if the RGC is an off centre or on centre cell?

• The boundary between a high (white) and low (black) response is the other way round in the output