the visual system v neuronal codes in the visual system

The visual system

VNeuronal codes in the

visual system

time

What‘s the code?

Firing rate Spike timing

- Synchrony- Timing patterns

’Firing rates are the only code that ALWAYS works’

The codes – firing rate

We start with the question

Does the brain use rate or precise timing?

We turn that into:

How noisy are networks?

The codes – firing rate

Latham & London (submitted)

Identical inputon every trial

t=0

The codes – firing rate

Latham & London (submitted)

large noise

one extra spike on trial 2

small noise

t=0

Identical inputon every trial

Latham & London (submitted)

We start with the question

Does the brain use rate or precise timing?

We turn that into:

How noisy are networks?

And finally:

How many extra postsynaptic spikes arecaused by one extra presynaptic spike?

The codes – firing rate

Latham & London (submitted)

Experimental details:

• in vivo whole cell recordings

• layer 5 pyramidal cells of rat barrel cortex

• urethane anesthetic

• with and without whisker stimulation

• current injection rather than PSPs

Latham & London (submitted)

V

100 ms

θ

Latham & London (submitted)

V

100 ms

θ

extra spike

Latham & London (submitted)

V

100 ms

θ

small effect

Latham & London (submitted)

V

100 ms

θ

Latham & London (submitted)

small effect

V

100 ms

θ

Latham & London (submitted)

big effect!!!

number of extra spikes caused by just one extra spike

= p1 × number of connections per neuron

≈ p1 × 1000

≈ 0.025 × 1000

= 25

Latham & London (submitted)

large noise

one extra spike on trial 2

small noise

t=0

Identical inputon every trial

Latham & London (submitted)

Manipulation of firing rates influences visual perception

Salzman et al., (1992)

Manipulation of firing rates influences visual perception

Salzman et al., (1992)

The codes – synchrony

’Perception is about association. Synchrony is too.’

The codes – synchrony

The codes – synchrony

The codes – synchrony

Center-surround interactions

Biederlack et al. (2006)

Center-surround interactions

Biederlack et al. (2006)

The escape of the bullfrog

Ishikane et al. (2005)

The escape of the bullfrog

Ishikane et al. (2005)

The codes – precise timing

’If it works, precise timing has incredible coding capacity’

20 ms per stage!

1 spike per neuron!

Thorpe & Fabre-Thorpe (2001)

The codes – precise timing

20-40 ms

30-50 ms40-50 ms

50-70 ms

70-90 ms

80-100 ms

What can one spike tell us?

What can one spike tell us?

Theories on spike timing in the cortex

Van Rullen & Thorpe (2001)

Onset latencies in vision

Gollisch & Meister (2008)

Fast OFF cell Biphasic OFF cell

Time[ms] Time[ms]

Onset latencies in vision

Gollisch & Meister (2008)

From external to internal timing

Experimental setup

• Anaesthesia

• Primary visual

cortex

• Grating stimuli

• 16 channels per

recording probe

• Multi- and single

unit activity

0.2 mm

Raw data

Time [ms]

Neuro

n #

Raw data

Time [ms]

Neuro

n #

Raw data

Time [ms]

Neuro

n #

Raw data

Time [ms]

Neuro

n #

Preferred firing sequences

Preferred relative firing time [ms]

Stimulus-dependent changes

Relative firing time [ms]

Stability

Relative firing time [ms]

7 ho

urs

Firing sequences and firing rates

rtotal = 0.28

r2total = 0.08

Firing rate

Firing time

Firing sequences and firing rates

Time [sec]

# of

act

ion

pote

ntia

ls

Rel

ativ

e fir

ing

time

[ms]

Time [sec]

rtotal = 0.01

r2total = 0.00

Neuronal coding in the real world

– what is a response?

Responses are multi-dimensional

Basole et al. (2003)

Information from ‘non-responsive‘ areas

Haxby et al. (2001)

Natural vision is dynamic

Things move.The body moves.Your eyes move.

Everything moves.

Vision is made to be a dynamic process.

´Lab´ activation

Mainen & Sejnowski (1995)

´Natural´ activation

Mainen & Sejnowski (1995)

Retinal responses to dynamic stimuli

Meister & Berry (1999)

The fly in the woods

Lewen et al. (2001)

The fly in the woods

Lewen et al. (2001)

Time (sec)

Sparse responses in natural vision

What‘s the

code?!

Neuronal coding in the real world –

what is a signal?

Strength and structure of inputs complement each other• Synaptic efficacy is boosted by bursting

of a single neuron and synchrony of several neurons (Usrey et al.,1998, 2000; Swadlow & Gusev, 2001)

• Integration time of retinal and LGN cells changes from 1 ms to 100 ms depending on visual circumstances (Berry & Meister 1999, Butts & Stanley, 2007)

Rall (1964)

Strength and structure of inputs complement each other

Rall (1964)

Strength and structure of inputs complement each other

Rall (1964)

Strength and structure of inputs complement each other

Rall (1964)

Strength and structure of inputs complement each other

Rall (1964)

Strength and structure of inputs complement each other

Rall (1964)

Strength and structure of inputs complement each other

Rall (1964)

Strength and structure of inputs complement each other

Euler & Denk (2004) Stiefel & Sejnowski (2007)

Strength and structure of inputs complement each other

Inputs modulate both rate and timing

Kuffler (1953)

Incr

ease

in s

tim

ulu

s in

tensi

ty

Stimulus onset50 ms

Inputs modulate both rate and timing

Fries et al. (2007)

Input

Input

Inputs modulate both rate and timing

Lengyel et al. (2005) Stiefel et al. (2005)

Summary V – Neuronal codes in the visual system…

• are often brought into conceptual competition

although in every day vision, they coexist naturally

• can rarely be tested directly to find out whether

they are crucial for perception

• are diverse and have all proven successful in

different visual tasks and circumstances

The code is…

Everything.