What we expect makes our visual system go PING!

What we expect to see, and what our brain predicts changes how our visual system reacts even when no visual information is present.

You’re probably wondering how this all works? For this we’re going to have to go down to the lab and check out a recent study done by the group of Floris de Lange (Donders Institute). One way to find out whether (a.) our brain generates some predictions about the world (let’s call it a template) and (b.) to find out more about the content of that template is to look at people’s brain activity in a task where we know for sure that people are going to generate predictions and under which we have a good idea of what the predictions will be made of. Ready?

  1. Bring some people into your lab.
  2. Give your participants a simple but effective task, which consists in generating strong expectations.

– For example, play a tone (let’s call it Tone A) so that it signals that a grating (series of lines) of a specific orientation is about to come and play another tone (Tone B) to signal another orientation.
– Make sure this cue is reliable (100% predictive) but make sure to introduce trials during which the visual stimuli is omitted.
– Make sure these trials are rare overall. These are your critical trials!
3. Record their brain activity with an MRI scanner (so you know where in the brain the activity is produced).
4. Expect the following:
– If the brain does generate strong predictions, it will be “surprised” on the critical trials. Indeed, it expected something to come on the screen but nothing came. This, we known, is likely to result in increased brain activity.
– If the brain generates predictions that are like a template then features that are specific of the stimulus that was expected should be visible.
This is exactly what Peter Kok, Michel Failing and Floris de Lange did. Being a participant in that study required you to judge whether two subsequently presented gratings (see figure 2) were of the same orientation or not. Before each series of gratings were presented, participants would hear a tone (tone A or tone B) which indicated with 100% reliability whether the grating coming after was oriented to the right (tone A, see figure 1) or to the left (tone B, see figure 1).
On 25% of the trials, however, the gratings were purposefully omitted.

Kok_etal_cueFig
Figure 1. This figure depicts the visual gratings associated to the two tones. ^1

Kok_etal_trialFigure
Figure 2. This figure depicts the two types of trials used in the experiment.^1

Did the brain of their participants react to the omission?
You bet! But first let’s quickly see what happened on the trials where the expected gratings actually came up. Unsurprisingly, because their visual system had some work to do, namely, processing the fact that something came up on the screen we can see a typical peak of activity circa 6 seconds after stimulus onset (green line see Figure 3). So far so good!

But look at the red line in the following figure. This line plots the amount of activity in response to the cases where participants were expecting to see a grating appear but did not. What is striking is that this line did not stay flat. This means that the visual cortex of the participants responded as if something had come up on the screen, as if it had to do some work to parse visual stimulus although, in fact, nothing was present on the screen.

Kok_etal_V1activations
Figure 3. Signal evoked by the presence or omission of gratings from participants visual cortex (V1). ^2

NOTE: If you don’t know how fMRI works you might feel a little lost here. Check out this companion post I wrote especially for you.

Ok, but was their brain just expecting something or was it expecting something particular?
Not all neurones react to stimuli the same way. In early visual cortices (V1) some neurones only care about some features while other care about other features. In the present case, some neurones only react when they see lines oriented at 45° and some only react when they see lines oriented at 135°.
What the researchers saw was that not only did the brain of participants fire even in the absence of expected stimuli but only the population of neurones typical of the expected orientation fired. So for example, after Tone A (see Figure 1) those neurones that like 45° orientations fired more than those that like 135° and vice-versa for Tone B.

Expectations, which are most probably generated by areas of the brain much later than visual ones, can activate visual features typical of what is expected, so that stimuli that is expected gets processed more easily and efficiently and only in the case of unexpected stimuli will our brain devote full computational power.

— References
ResearchBlogging.org
Kok, P., Failing, M., & de Lange, F. (2014). Prior Expectations Evoke Stimulus Templates in the Primary Visual Cortex Journal of Cognitive Neuroscience, 26 (7), 1546-1554 DOI: 10.1162/jocn_a_00562
— Footnotes
^1. Figure taken from Kok et al., 2014
^2. Figure adapted from from Kok et al., 2014. Original can be seen in the article referenced above.

Trackbacks for this post

  1. Words jump-start our visual system | bastienboutonnet.com | research & blogging

Leave a Reply