1 of 37

The neural code�Single-cell temporal coding

Kenneth D Harris, UCL

2 of 37

“Temporal code”

Not just firing rate, but precise spike timing conveys information

In principle, neurons could convey much more information using precise timing

Does it really happen?

Example: brain transforms information into temporal code, and communicates it through the air to another brain.

(spoken language)

Does this also happen for communication within a single brain?

3 of 37

Outline

Definition of temporal coding

Examples from the literature

Exploratory analyses for temporal codes

International Brain Lab data

Confirmatory analyses of temporal coding

Allen Institute data

4 of 37

Outline

Definition of temporal coding

Examples from the literature

Exploratory analyses for temporal codes

International Brain Lab data

Confirmatory analyses of temporal coding

Allen Institute data

5 of 37

Temporal coding in the endocrine system

Feminization of Male Mouse Liver by Persistent Growth Hormone Stimulation: Activation of Sex-Biased Transcriptional Networks and Dynamic Changes in Chromatin States.

Lau-Corona et al. Molecular and Cellular Biology 2017

Somatostatin Is Essential for the Sexual Dimorphism of GH Secretion, Corticosteroid-Binding Globulin Production, and Corticosterone Levels in Mice.

Adams et al, Endocrinology 2015

6 of 37

3 Types of temporal code

1: Following the timing of an external stimulus

2: Sequence of activity reflecting steps in a serial computation

3: Generating new temporal structure that encodes non-temporal information

Plus:

0: decoding a temporally structured external stimulus into a rate-coded pattern

7 of 37

Type 1 temporal coding : following stimulus timing

8 of 37

Type 2 temporal coding : sequential computation

Hebb’s “phase sequence”: a sequence of assemblies implementing a serial computation

Monkey expression

Human identity

Human, monkey or shape?

9 of 37

A related idea: binding by synchrony

Each step of the phase sequence represents a coherent set of features

Could the brain “multiplex” the features of different objects at different phases of an oscillation?

10 of 37

Type 3 temporal coding: generating timing internally

Responses to tones, auditory cortex
Responses to preferred tones are slightly earlier
Differences between neurons greater than differences between stimuli

11 of 37

Type 0: temporal decoding

12 of 37

Outline

Definition of temporal coding

Examples from the literature

Exploratory analyses for temporal codes

International Brain Lab data

Confirmatory analyses of temporal coding

Allen Institute data

13 of 37

International Brain Lab data

621733 neurons total

14 of 37

Raster plot and peristimulus time histogram (PSTH)

Experiment ID:

e0928e11-2b86-4387-a203-80c77fab5d52

Unit 45, Superior Colliculus intermediate layers

15 of 37

Sorting a raster plot

16 of 37

Sorting a raster plot

17 of 37

Sorting a raster plot

18 of 37

Sorting a raster plot

19 of 37

Sorting a raster plot

20 of 37

Sorting a raster plot

21 of 37

Dissociate colors from sorting

22 of 37

Dissociate colors from sorting

23 of 37

Sorting by response time

Time of first movement

Colored by right contrast

24 of 37

Aligning to response time

Time of stimulus onset

Colored by reaction time

Perimovement time

25 of 37

Fine structure of spike trains

Superior colliculus cells in the first 3s of the experiment. (Task has not started)

26 of 37

Interspike interval histogram and Autocorrelogram

Refractory period

Mean firing rate

Burst shoulders

27 of 37

Different cell types can have different autocorrelograms

Example: in the striatum

Medium spiny neurons (MSNs): inhibitory projection neurons
Fast-spiking interneurons: inhibitory local neurons
Tonically-active neurons: cholinergic local neurons

Have different autocorrelograms that you can use to distinguish them
Most brain regions: uknown!

Spike waveform

Autocorrelogram

28 of 37

Outline

Definition of temporal coding

Examples from the literature

Exploratory analyses for temporal codes

International Brain Lab data

Confirmatory analyses of temporal coding

Allen Institute data

29 of 37

Allen Institute mouse visual cortex data

Problem with task data: if a neuron encodes both stimulus and response, and response time depends on stimulus contrast, how do you know it is coding the stimulus specifically?
We’ll deal with that later
For now just use passive stimulus responses

Siegle et al Nature 2021. https://www.nature.com/articles/s41586-020-03171-x

30 of 37

Allen ephys session 791319847: unit 180 (primary visual cortex)

Null hypothesis: spike counts may differ between stimuli, but spike timing doesn’t
To test: shuffle peristimulus times between spikes, keeping trial/stimulus allocations the same

Original spike times

31 of 37

Allen ephys session 791319847: unit 180 (primary visual cortex)

Null hypothesis: spike counts may differ between stimuli, but spike timing doesn’t
To test: shuffle peristimulus times between spikes, keeping trial/stimulus allocations the same

Shuffled spike times

32 of 37

Allen ephys session 791319847: unit 181 (primary visual cortex)

Null hypothesis: spike counts may differ between stimuli, but spike timing doesn’t
To test: shuffle peristimulus times between spikes, keeping trial/stimulus allocations the same

Original spike times

33 of 37

Allen ephys session 791319847: unit 181 (primary visual cortex)

Null hypothesis: spike counts may differ between stimuli, but spike timing doesn’t
To test: shuffle peristimulus times between spikes, keeping trial/stimulus allocations the same

Shuffled spike times

34 of 37

Allen ephys session 791319847: unit 0 (dentate gyrus)

Null hypothesis: spike counts may differ between stimuli, but spike timing doesn’t
To test: shuffle peristimulus times between spikes, keeping trial/stimulus allocations the same

Original spike times

35 of 37

Allen ephys session 791319847: unit 0 (primary visual cortex)

Null hypothesis: spike counts may differ between stimuli, but spike timing doesn’t
To test: shuffle peristimulus times between spikes, keeping trial/stimulus allocations the same

Shuffled spike times

36 of 37

Permutation test p-values

Test statistic: std deviation of mean spike times, across stimuli

Cell 180: p=0.001

Cell 181: p=0.001

Cell 0: p=0.893

1 of 37

2 of 37

3 of 37

4 of 37

5 of 37

6 of 37

7 of 37

8 of 37

9 of 37

10 of 37

11 of 37

12 of 37

13 of 37

14 of 37

15 of 37

16 of 37

17 of 37

18 of 37

19 of 37

20 of 37

21 of 37

22 of 37

23 of 37

24 of 37

25 of 37

26 of 37

27 of 37

28 of 37

29 of 37

30 of 37

31 of 37

32 of 37

33 of 37

34 of 37

35 of 37

36 of 37

37 of 37