Virtual

ERP Boot Camp

Steve Luck Emily Kappenman
Center for Mind & Brain Department of Psychology
University of California, Davis San Diego State University

ERP Basics
These materials were made possible by NIH grant R25MH080794 and are shared
under the terms of a Creative Commons license (CC BY-SA 4.0)
 Virtual
ERP Boot Camp
Steve Luck Emily Kappenman
Center for Mind & Brain Department of Psychology
University of California, Davis San Diego State University

ERP Basics
The EEG

These materials were made possible by NIH grant R25MH080794 and is

shared under the terms of a Creative Commons license (CC BY-SA 4.0)
The Electroencephalogram (EEG)

The electrodes don’t directly

contact the skin. Instead, we squirt
in a conductive gel that makes
contact between the skin and the
metal electrode pellet.
 Notice the oscillations at
these electrode sites. If you
counted, you’d find about 10
peaks in a one-second
period. That means it’s a 10
Hz oscillation.
 Major EEG Bands
Raw EEG
0.0 0.2 0.4 0.6 0.8 1.0

Delta (1-3 Hz) Slow-wave sleep

0.0 0.2 0.4 0.6 0.8 1.0

Not the same as

Theta (4-7 Hz) hippocampal theta
0.0 0.2 0.4 0.6 0.8 1.0

Internal rather than

Alpha (8-12 Hz)
external focus
0.0 0.2 0.4 0.6 0.8 1.0

Beta (12-30 Hz) Mentally active

0.0 0.2 0.4 0.6 0.8 1.0

Local
Gamma (30+ Hz)
communication
0.0 0.2 0.4 0.6 0.8 1.0

https://en.wikipedia.org/wiki/Eeg
 Virtual
ERP Boot Camp
Steve Luck Emily Kappenman
Center for Mind & Brain Department of Psychology
University of California, Davis San Diego State University

ERP Basics
Averaged ERPs

This video was made possible by NIH grant

R25MH080794 and is shared under the terms of a
Creative Commons license (CC BY-SA 4.0)
Event-Related Potentials Task-Dependent Emotional Response
Stimulus
Electrical potentials (voltages) Categorization
Initial
that are related to specific Sensory Response
events Response Selection
Higher Working
Perceptual Memory
Processes Encoding

P3
P1 P2
+1 µV
N2

-1 µV
C1 N1

0 200 400 600 ms

 Stimulation
Computer Event Codes
Event Codes

Filter, Amplify,
Digitize

Most labs use two computers, one

that presents the stimuli and records
the responses, and another that
records the EEG and event codes
 Stim 1 Stim 2 Stim N

To pull out the brain’s

Voltage
consist response to some
Time type of event, we can simply
EEG Epochs average across the epochs
Averaged ERP
+ Waveform for that event type.
Stim 1 P3
When we average across
P2 enough epochs, any activity
P1
Stim 2 µV
that’s consistent from trial to
N2 trial remains in the average,
N1
Stim N – and any random noise
0 200 400 600
simply averages out.
Time (ms)
Stimulus Onset 600 ms Stimulus Onset
 In a typical ERP Task-Dependent Emotional Response
Stimulus
experiment, we time-lock to Initial
Categorization

the onset of a stimulus. In Sensory Response

Response Selection
other words, time zero is
Higher Working
stimulus onset. Perceptual Memory
Processes Encoding
When we do this, we can
track the flow of information
P3
about the stimulus through
the brain over a period of P1 P2
several hundred +1 µV
milliseconds. N2

-1 µV
C1 N1

0 200 400 600 ms

 Mean Response Time Although most
(relative to stimulus onset) researchers focus on
Stimulus-Locked Average stimulus-locked
averages, where time
zero is stimulus onset,
-200 0 200 400 600 you can instead time-
Stimulus Onset
lock to the response.
This gives you a
Mean Stimulus Time response-locked
(relative to response) Response-Locked Average average.

-600 -400 -200 0 200

Response
 Virtual
ERP Boot Camp
Steve Luck Emily Kappenman
Center for Mind & Brain Department of Psychology
University of California, Davis San Diego State University

ERP Basics
Example: The N170 Component
and Perceptual Experience
This video was made possible by NIH grant
R25MH080794 and is shared under the terms of a
Creative Commons license (CC BY-SA 4.0)
Tanaka, J. W., & Curran, T. (2001). A neural basis for expert
object recognition. Psychological Science, 12, 43–47.
 N170 and Face Processing
Stimuli Grand Average ERPs

Cars

Faces
The waveforms shown here are averaged
across all the subjects – we call these grand
averages.
The N170 wave was much bigger for the
faces than for the cars.
Rossion & Jacques (2012)
 Is face perception special?

Do we have domain-specific neural systems that are solely used for face
perception? That might explain why inverting an image has a larger impact on
the perceptibility of faces than on other sorts of objects. But maybe these
effects are a consequence of the fact that virtually all humans have a lifetime of
expertise in perceiving faces.
 Is face perception special?

Or do we use the same processes for all

stimuli that we’re experts at perceiving?
 Paradigm and Predicted Results
Bird Experts Dog Experts
Bird Stimuli Large N170 Small N170
Dog Stimuli Small N170 Large N170

Factor 1: Expertise of participant (Bird vs. Dog)

Factor 2: Stimulus type (Bird vs. Dog)

Robin True/False

Fixation Category Name Fixation Picture Fixation Response

1000-1500 ms 255 ms 570 ms 255 ms 735 ms ???
The N170 was larger for bird pictures than for dog pictures in the bird experts.
But N170 was larger for dog pictures than for bird pictures in the dog experts.
This is consistent with the idea that face perception is achieved by a set of
general-purpose processes that are not face specific but depend on expertise.

Tanaka & Curran (2001)

Rossion, B., & Jacques, C. (2012). The N170: Understanding the time course of face perception in the human brain. In S.
J. Luck & E. S. Kappenman (Eds.), The Oxford Handbook of ERP Components (pp. 115–141). Oxford University Press.
 Virtual
ERP Boot Camp
Steve Luck Emily Kappenman
Center for Mind & Brain Department of Psychology
University of California, Davis San Diego State University

ERP Basics
The Signal-to-Noise Ratio

This video was made possible by NIH grant

R25MH080794 and is shared under the terms of a
Creative Commons license (CC BY-SA 4.0)
How many trials do you need to average together?

on the size of the signal relative to the

size of the noise
(the signal-to-noise ratio [SNR])
Oddball Paradigm: X X X X O X X X O X X
Oddball Oddball

Trial 1 Trial 5
EEG EEG

Trial 2 Trial 6
EEG EEG

Trial 3 Trial 7
EEG EEG

Trial 4 Trial 8
EEG EEG

You can see a lot of variation from trial to trial, but all 8 trials have a positive
+15µV
Average of
voltage from approximately
8 Trials 300 to 600 ms. That’s the P3 wave, which is a very
-200 200 400 600 800 ms
large positive component that you get for oddballs in this paradigm.
–15µV
Oddball Paradigm: X X X X O X X X O X X
Oddball Oddball

Trial 1 Trial 5
EEG EEG

Trial 2 Trial 6
EEG EEG

Trial 3 Trial 7
EEG EEG

Trial 4 Trial 8
EEG EEG

P3
+15µV
Average
Average of 8 Oddball Trialsof
8 Trials
-200 200 400 600 800 ms
–15µV
Oddball Paradigm: X X X X O X X X O X X
Oddball Oddball

Trial 1 Trial 5
EEG EEG

Trial 2 Trial 6
EEG EEG

Trial 3 Trial 7
EEG EEG

Trial 4 Trial 8
EEG EEG

P3
+15µV
Average
Average of 8 Oddball Trialsof
8 Trials
-200 200 400 600 800 ms
–15µV
Oddball Paradigm: X X X X O X X X O X X
Oddball Oddball

Trial 1 Trial 5
EEG EEG

Trial 2 Trial 6
EEG EEG

If the signal isn’t any bigger than the

Trial 3 noise, it’s going toTrialbe
7 hard to find
EEG EEG
statistically significant differences
between conditions or groups
Trial 4 Trial 8
EEG EEG

P3
+15µV
P2
Average
Average of 8 Oddball Trialsof
8 Trials
-200 200 400 600 800 ms
–15µV
Oddball Paradigm: X X X X O X X X O X X
Oddball Oddball

Trial 5
Trial 1
EEG Increasing the number EEGof trials doesn’t
change the signal – it just decreases
the noise. So, if we quadrupled the
Trial 2 Trial 6
EEG number of trials to double
EEG the SNR,
this would effectively cut the noise in
Trial 3 half.Trial 7
EEG EEG

Signal-to-noise ratio a sqrt(number of trials)

Trial 4 Trial 8
EEG EEG

P3
+15µV
P2
Average
Average of 8 Oddball Trialsof
8 Trials
-200 200 400 600 800 ms
–15µV
 Virtual
ERP Boot Camp
Steve Luck Emily Kappenman
Center for Mind & Brain Department of Psychology
University of California, Davis San Diego State University

ERP Basics
Sources of Noise

This video was made possible by NIH grant

R25MH080794 and is shared under the terms of a
Creative Commons license (CC BY-SA 4.0)
 Where does the noise come from?
1: Brain activity that is not time-locked to the event of interest

Trial 1 Trial 5
EEG EEG

Trial 2 Trial 6
EEG EEG

Trial 3 Trial 7
EEG EEG

Trial 4 Trial 8
EEG EEG

+15µV
Average of
8 Trials
-200 200 40
–15µV
Where does the noise come from?
2: Biological artifacts
Electrooculogram

Electromyogram Skin Potentials

Hair

Epidermis
Skin
Surface

Sebum Sweat
Pore Dermis

Sebaceous
Gland

Sweat
Gland
 Where does the noise come from?
3: Induced electrical activity from the recording environment
Magnetic field
generated around
conductor #1

Current flows Conductor #1

through conductor #1
Conductor #2

Magnetic field
induces current flow
in conductor #2
 Where does the noise come from?
What do we mean by noise, anyway?
Noise: any source of uncontrolled
variability in the signal of interest
Hair

Epidermis
Skin
Surface

Sebum Sweat
Pore Dermis

Sebaceous
Gland

Sweat
Gland
Trial 1 Trial 5
EEG EEG

Trial 2 Trial 6
EEG EEG

Trial 3 Trial 7
EEG EEG

Trial 4 Trial 8
EEG EEG
 Where does the noise come from?
What do we mean by noise, anyway?
Noise: any source of uncontrolled
variability in the signal of interest

Trial 1 Trial 5
EEG EEG

Trial 2 Trial 6
EEG EEG

Trial 3 Trial 7
EEG EEG

Trial 4 Trial 8
EEG EEG

+15µV
Average of
8 Trials
-200 200 400 600
–15µV
 Virtual
ERP Boot Camp
Steve Luck Emily Kappenman
Center for Mind & Brain Department of Psychology
University of California, Davis San Diego State University

ERP Basics
Common Conventions in
ERP Research
This video was made possible by NIH grant
R25MH080794 and is shared under the terms of a
Creative Commons license (CC BY-SA 4.0)
The International 10/20 System

From Malmivuo, J. & Plonsey, R. (1995)

 Front
The International 10/20 System Nz

Fp1 Fpz Fp2

AF7 AF8
AF3 AF4
Fp: Frontal Pole F9
AFz
F10

F: Frontal
F7 F8
F5 F6
F3 F1 Fz F2 F4

C: Central (Vertex) Left FT9 FT7 FT8 FT10

Right
FC5 FC6
P: Parietal
FC3 FC1 FCz FC2 FC4
Ear Ear
O: Occipital A1 T9 T7 C5 C3 C1 Cz C2 C4 C6 T8 T10 A2
T: Temporal
CP5 CP3 CP1 CPz CP2 CP4 CP6
TP7 TP8
TP9 TP10
P1 Pz P2
Left: Odd Numbers P5 P3 P4 P6
P7 P8

Right: Even Numbers

P9 P10
PO3 POz PO4
PO7 PO8
Midline: z for zero O1 O2
Oz

Iz

Back
 Front
The International 10/20 System Nz

Fp1 Fpz Fp2

AF7 AF8
AF3 AFz AF4
F9 F10
F7 F8
F5 F6
F3 F1 Fz F2 F4

Left FT9 FT7 FC5 FC3 FC1 FCz FC2 FC4 FC6 FT8 FT10
Right
Ear Ear
A1 T9 T7 C5 C3 C1 Cz C2 C4 C6 T8 T10 A2

CP5 CP3 CP1 CPz CP2 CP4 CP6

TP7 TP8
TP9 TP10

P3 P1 Pz P2 P4
Alternative: Geodesic P7
P5 P6
P8
electrode configuration P9 P10
PO3 POz PO4
PO7 PO8
Geodesic layouts use idiosyncratic numbering O1 O2
Oz
systems, and a publication will typically say
something like “These are data from electrode
Iz
27, which near the P2 location in the
International 10/20 system. Back
-200 0 200 400 600
 5 FP1
3.8
2.5
5 FZ 5 FZ
3.8
2.5
5 F3
3.8
2.5
5 F7
3.8
2.5
5 F3 5 T7
3.8
2.5
5 C3
3.8
2.5
5
1.3

-100
-1.3
3.8
100 200 300 400
1.3

-100
-1.3 100 200 300 400
1.3

-100
-1.3 100 200 300 400
1.3

-100
-1.3 100 200 300 400
3.8 1.3

-100
-1.3 100 200 300 400
1.3

-100
-1.3 100 200 300 400
3.8
-2.5
2.5 -2.5 -2.5 -2.5
2.5 -2.5 -2.5
2.5
We record 1.3
from
many electrode sites But the simultaneously.
-3.8 -3.8 -3.8 -3.8 -3.8 -3.8

5 CZ 5 P1 5 P3 5 P5
1.3 5 P7 5 P9
1.3
3.8 3.8 3.8 3.8 3.8 3.8

waveforms at nearby sites are typically quite similar, so there’s usually

2.5
1.3
2.5
1.3
2.5
1.3
2.5
1.3
2.5
1.3
2.5
1.3

100 200 300 400

-100
-1.3 -100
-1.3 100 200 300 400
100 200 300 400 -100
-1.3 -100
-1.3 100 200 300 400
100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3
no point in showing
-2.5
-3.8
-2.5 all of the electrode sites
-2.5
-3.8
-2.5 in a journal article.
-2.5
-3.8
-2.5
-3.8
-2.5
-3.8
-2.5
-3.8
-2.5
5 PO7
3.8
2.5
-3.8
5 PO3
3.8
2.5
5 O1
3.8
2.5
5 IZ
3.8
2.5
-3.8 5 FP2
3.8
2.5
5 F4
3.8
2.5
-3.8
1.3 1.3 1.3 1.3 1.3 1.3

-100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400
-2.5 -2.5 -2.5 -2.5 -2.5 -2.5

5 P1 5 P3
-3.8 -3.8 -3.8 -3.8 -3.8 -3.8

5 F8 5 T8 5 C4 5 PZ 5 P2 5 P4
5
3.8
2.5
3.8
2.53.8 3.8
2.5
3.8
2.5 3.8 3.8
2.5
3.8
2.5 3.8
1.3 1.3 1.3 1.3 1.3 1.3
-100
-1.3 100 200 300 400 -100
2.5
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400
2.5
-100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400
2.5
-2.5
-3.8 1.3
-2.5
-3.8
-2.5
-3.8
-2.5
-3.8 1.3 -2.5
-3.8
-2.5
-3.8 1.3
5 P6 5 P8 5 P10 5 PO8 5 PO4 5 POZ
3.8 3.8 3.8 3.8 3.8 3.8
100 200 300 400 2.5
1.3 -100
-1.3
2.5
1.3 100 200 300 400
2.5
1.3
2.5
1.3 -100
-1.3 2.5
1.3 100 200 300 400 2.5
1.3 -100
-1.3
-100
-1.3
-2.5
100 200 300 400 -2.5
-100
-1.3
-2.5
100 200 300 400 -100
-1.3
-2.5
100 200 300 400 -100
-1.3
-2.5
100 200 300 400 -2.5
-100
-1.3
-2.5
100 200 300 400 -100
-1.3
-2.5
100 200 300 400 -2.5
-3.8
-3.8
-3.8 -3.8 -3.8
-3.8 -3.8 -3.8
-3.8
5 OZ 5 O2 5 M1 5 M2 5 HEOG 5 VEOG
3.8 3.8 3.8 3.8 3.8 3.8
2.5 2.5 2.5 2.5 2.5 2.5
1.3 1.3 1.3 1.3 1.3 1.3

5 PO3 5 O1
-100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400
-2.5
-3.8
-2.5
-3.8
-2.5
-3.8
-2.5
-3.8
-2.5
-3.8
-2.5
-3.8 5
3.8 3.8 3.8
2.5 2.5 2.5
Simultaneous
1.3 recordings from 1.3 1.3
many electrode sites
100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3 100 200 300 400 -100
-1.3
-2.5 -2.5 -2.5
-3.8 -3.8 -3.8

5 T8 5 C4 5
 These maps use interpolation to show
voltage over the entire scalp, even
though we only have measures at the
discrete electrode sites.

100 ms 150 ms 200 ms 250 ms 300 ms

Subject 1

Subject 2 ERP papers don’t usually show

single-subject ERP waveforms.
Subject 3
Instead, we usually take the single-
Subject 4 subject averaged ERP waveforms and
average them together into a Grand
Subject 5
Average waveform.
Subject 6
However, the statistical analysis is
Subject 7
done using the single-subject
waveforms, not the grand
+5µV
Subject 8 averages.
–5µV 200 400 600 800 ms

+5µV
Grand
Average 200 400 600 800 ms
–5µV
 This study measured the mean
amplitude in the N170 time range
for each individual subject’s
averaged ERP waveform in each
condition and then put those
single-subject amplitude values
into a simple ANOVA.

Tanaka & Curran (2001)

 2005 2014

Plotting negative-up is an outdated convention that arose from a

historical accident. It’s still pretty common, so you’ll need to make sure
you look at the polarity indicators in plots of ERP waveforms.

-5 μV
-200 200 400 600
Maximum Voltage

Local Peak
+5μV
+5 μV
Local Peak
Maximum Voltage
-200 200 400 600
-5μV