WEBVTT

00:00:00.000 --> 00:00:02.520 align:middle line:84%
So Laura, we both do
pupillometry research.

00:00:02.520 --> 00:00:05.230 align:middle line:84%
But we use it in
very different ways.

00:00:05.230 --> 00:00:09.990 align:middle line:84%
So my own work focuses more on
these evoked pupil responses,

00:00:09.990 --> 00:00:14.160 align:middle line:84%
happen on a much smaller,
quicker, time scale in the lab.

00:00:14.160 --> 00:00:17.730 align:middle line:84%
Whereas you work more with
steady-state changes, overall,

00:00:17.730 --> 00:00:18.930 align:middle line:90%
in the pupil, right?

00:00:18.930 --> 00:00:20.160 align:middle line:90%
Most of the time, yeah.

00:00:20.160 --> 00:00:22.800 align:middle line:84%
So this means that
I am capturing

00:00:22.800 --> 00:00:26.130 align:middle line:84%
pupil data over a period
of minutes, for example.

00:00:26.130 --> 00:00:28.980 align:middle line:84%
And then taking an
average of the pupil

00:00:28.980 --> 00:00:31.590 align:middle line:84%
size across that
period and comparing it

00:00:31.590 --> 00:00:35.890 align:middle line:84%
to the average pupil size that's
captured several minutes later.

00:00:35.890 --> 00:00:38.760 align:middle line:84%
So these are
capturing more slowly,

00:00:38.760 --> 00:00:42.300 align:middle line:84%
unfolding changes that are in
response to a condition that

00:00:42.300 --> 00:00:44.620 align:middle line:84%
endures for a longer
period of time.

00:00:44.620 --> 00:00:46.210 align:middle line:90%
Great, really interesting stuff.

00:00:46.210 --> 00:00:48.480 align:middle line:84%
And so what kind
of things are you

00:00:48.480 --> 00:00:51.540 align:middle line:84%
looking to capture with
these longer time windows?

00:00:51.540 --> 00:00:54.900 align:middle line:84%
Well, since I'm usually working
with performing musicians,

00:00:54.900 --> 00:00:58.860 align:middle line:84%
I'm usually looking at how
the musicians are responding

00:00:58.860 --> 00:01:02.220 align:middle line:84%
with their attention to a
particular playing condition

00:01:02.220 --> 00:01:05.040 align:middle line:90%
that I have set up.

00:01:05.040 --> 00:01:07.860 align:middle line:84%
For example, I might
have them playing

00:01:07.860 --> 00:01:10.320 align:middle line:84%
for an audience
for several minutes

00:01:10.320 --> 00:01:12.750 align:middle line:84%
and compare this to
their attention states

00:01:12.750 --> 00:01:15.180 align:middle line:90%
when they're rehearsing.

00:01:15.180 --> 00:01:17.880 align:middle line:84%
So I think this is quite
different from what you're

00:01:17.880 --> 00:01:20.220 align:middle line:84%
doing because I'm
usually working

00:01:20.220 --> 00:01:23.490 align:middle line:84%
in quite an ecological
situation where

00:01:23.490 --> 00:01:25.440 align:middle line:84%
I have to preserve
the musician's

00:01:25.440 --> 00:01:26.430 align:middle line:90%
normal way of behaving.

00:01:26.430 --> 00:01:30.570 align:middle line:84%
But I think you're looking at
actually a lot more detailed

00:01:30.570 --> 00:01:31.300 align:middle line:90%
responses.

00:01:31.300 --> 00:01:34.890 align:middle line:84%
So you cannot capture this
kind of data out in the wild

00:01:34.890 --> 00:01:36.160 align:middle line:90%
in the real world.

00:01:36.160 --> 00:01:40.140 align:middle line:84%
Yeah, as much as I'd like to,
I do have to bring people in

00:01:40.140 --> 00:01:44.157 align:middle line:84%
and subject them to laboratory
conditions which, so far,

00:01:44.157 --> 00:01:46.740 align:middle line:84%
I can proudly announce only one
person is fallen asleep on me.

00:01:46.740 --> 00:01:48.960 align:middle line:90%
That's not too bad.

00:01:48.960 --> 00:01:50.550 align:middle line:84%
Yeah, it's a totally
different way

00:01:50.550 --> 00:01:52.230 align:middle line:90%
of investigating these things.

00:01:52.230 --> 00:01:55.307 align:middle line:84%
Because I think for
these small changes,

00:01:55.307 --> 00:01:57.390 align:middle line:84%
you really have to have
very controlled conditions

00:01:57.390 --> 00:01:59.460 align:middle line:90%
if you want to see the effect.

00:01:59.460 --> 00:02:02.190 align:middle line:84%
And can we talk a little bit
about what the process is

00:02:02.190 --> 00:02:06.900 align:middle line:84%
for going from where you have
just captured data to where

00:02:06.900 --> 00:02:08.729 align:middle line:84%
you're prepared to
analyse your data?

00:02:08.729 --> 00:02:12.600 align:middle line:84%
I think you have a demonstration
of what your data looks like

00:02:12.600 --> 00:02:14.790 align:middle line:90%
just after you've captured it.

00:02:14.790 --> 00:02:18.900 align:middle line:84%
Yeah, this is the output
that we get directly

00:02:18.900 --> 00:02:21.600 align:middle line:90%
from the eye tracker.

00:02:21.600 --> 00:02:26.040 align:middle line:84%
So I work with an EyeLink system
and it spits out these plots.

00:02:26.040 --> 00:02:28.800 align:middle line:84%
Personally, I try to
export it as early

00:02:28.800 --> 00:02:31.230 align:middle line:84%
as possible to something
that's a little bit more

00:02:31.230 --> 00:02:33.780 align:middle line:90%
user-friendly for me, anyway.

00:02:33.780 --> 00:02:35.100 align:middle line:90%
And so I've exported it.

00:02:35.100 --> 00:02:39.070 align:middle line:84%
I get something that
looks more like this.

00:02:39.070 --> 00:02:42.210 align:middle line:84%
And as you can see, we have
the subject number, some time

00:02:42.210 --> 00:02:43.320 align:middle line:90%
stamps--

00:02:43.320 --> 00:02:46.560 align:middle line:84%
which will take a little bit
of preprocessing, actually.

00:02:46.560 --> 00:02:48.930 align:middle line:84%
Well, after a little bit
of preprocessing magic

00:02:48.930 --> 00:02:54.200 align:middle line:84%
which everybody will learn about
sooner, or later this week,

00:02:54.200 --> 00:02:56.240 align:middle line:84%
then we start to
get prettier plots

00:02:56.240 --> 00:03:00.990 align:middle line:84%
like this, where I can show
the pupil trace over time.

00:03:00.990 --> 00:03:03.690 align:middle line:84%
And I also work with
averages, as well.

00:03:03.690 --> 00:03:06.120 align:middle line:84%
And so I've got that
showed here, as well.

00:03:06.120 --> 00:03:08.630 align:middle line:84%
So this is an average,
per condition,

00:03:08.630 --> 00:03:10.430 align:middle line:90%
across number of participants?

00:03:10.430 --> 00:03:12.830 align:middle line:84%
Yeah, so this is group
level averages now.

00:03:12.830 --> 00:03:17.960 align:middle line:84%
And you're looking at a rhythmic
complexity that's increasing

00:03:17.960 --> 00:03:21.800 align:middle line:84%
and the pupil will dilate with
increasing rhythmic complexity.

00:03:21.800 --> 00:03:25.130 align:middle line:84%
And this also is modulated
by whether or not people are

00:03:25.130 --> 00:03:28.880 align:middle line:84%
simply listening to it or if
they're tapping and listening.

00:03:28.880 --> 00:03:29.990 align:middle line:90%
This is very interesting.

00:03:29.990 --> 00:03:32.600 align:middle line:84%
Yeah, so what's your
data look like, actually?

00:03:32.600 --> 00:03:36.450 align:middle line:84%
Well, this is an example of
what some of my data looks like.

00:03:36.450 --> 00:03:41.150 align:middle line:84%
This is data, just
in the software

00:03:41.150 --> 00:03:45.480 align:middle line:84%
that is used to do the
initial stages of processing.

00:03:45.480 --> 00:03:49.160 align:middle line:84%
And so here you can
see the video, actually

00:03:49.160 --> 00:03:51.950 align:middle line:84%
that's captured from
the mobile eye trackers.

00:03:51.950 --> 00:03:53.990 align:middle line:84%
And you can see on
the bottom there's

00:03:53.990 --> 00:03:56.730 align:middle line:84%
some indication of
the pupil curve.

00:03:56.730 --> 00:03:58.610 align:middle line:90%
So this is the pupil data.

00:03:58.610 --> 00:04:01.640 align:middle line:84%
This is interesting
and useful to look

00:04:01.640 --> 00:04:03.410 align:middle line:84%
at the quality of
your data and see

00:04:03.410 --> 00:04:05.270 align:middle line:90%
what's going on with the gaze.

00:04:05.270 --> 00:04:07.190 align:middle line:84%
But, actually, I
would also prefer

00:04:07.190 --> 00:04:10.670 align:middle line:84%
to export my data
into a text format.

00:04:10.670 --> 00:04:14.095 align:middle line:84%
And mine looks quite
similar to yours.

00:04:14.095 --> 00:04:17.200 align:middle line:90%


00:04:17.200 --> 00:04:21.480 align:middle line:84%
So it's just a lot of time
stamps, participant numbers,

00:04:21.480 --> 00:04:24.090 align:middle line:90%
and pupil sizes.

00:04:24.090 --> 00:04:29.400 align:middle line:84%
And so after processing
this kind of data,

00:04:29.400 --> 00:04:34.870 align:middle line:84%
I would end up with plots that
look like this, for example.

00:04:34.870 --> 00:04:39.220 align:middle line:84%
So these are violin
and box plots together,

00:04:39.220 --> 00:04:40.970 align:middle line:84%
which I find is a
useful way of showing

00:04:40.970 --> 00:04:44.910 align:middle line:84%
the distribution of data
once you've aggregated it,

00:04:44.910 --> 00:04:48.940 align:middle line:84%
in this case, per condition
across many participants.

00:04:48.940 --> 00:04:52.140 align:middle line:84%
So you can see the way that
the data is distributed

00:04:52.140 --> 00:04:53.570 align:middle line:90%
within each condition.

00:04:53.570 --> 00:04:55.890 align:middle line:84%
Yeah, you can see that some
of them, like the grey one,

00:04:55.890 --> 00:04:58.020 align:middle line:84%
are quite skewed
towards the low end.

00:04:58.020 --> 00:04:59.160 align:middle line:90%
Yes, yes.

00:04:59.160 --> 00:05:01.770 align:middle line:84%
So this is also a
good way of checking

00:05:01.770 --> 00:05:04.320 align:middle line:84%
the quality of your data
and making sure you're not

00:05:04.320 --> 00:05:05.490 align:middle line:90%
missing anything.

00:05:05.490 --> 00:05:07.530 align:middle line:90%
Beautiful.

00:05:07.530 --> 00:05:09.120 align:middle line:90%
Now we want to hear from you.

00:05:09.120 --> 00:05:12.350 align:middle line:84%
How do you plan on
using pupil analysis?

00:05:12.350 --> 00:05:23.000 align:middle line:90%