Powerpoint - personal.rdg.ac.uk

download report

Transcript Powerpoint - personal.rdg.ac.uk

Timing of the brain events underlying access
to consciousness during the attentional blink
Claire Sergent, Sylvain Baillet, & Stanislas
Dehaene
The attentional blink
Perception of the first target (T1) impairs perception
of the second (T2) if the latter follows the former within
an interval of 500 ms (Raymond, Shapiro & Arnell (1992).
Principal question: what is the fate of blinked stimuli?
- Are blinked stimuli unconsciously perceived?
--- Yes. Vogel, Luck & Shapiro (1998). P3 suppression vs N4
- Can the difference between identical blinked and perceived stimuli be observed in the
event-related potentials?
--- Sergent et al. compare the following conditions:
T2 present, seen – T2 absent; T2 present, unseen – T2 absent; T2 seen – T2 unseen
Normally, Task2 requires some for of identification, but the present study uses
a visibility rating. However, this method has been validated in Sergent &
Dehaene (2004).
2
- The 50% visibility cut-off is used for the seen/unseen contrast.
- Task1 is a 2-AFC, which is fairly easy. Stimulus presentation of 43 ms makes up for it.
Results
FCz
C1
Un-subtracted waveforms from Suppl. Figure 1 (– T2 seen; – T2 unseen; -- T2 absent).
- These show the evoked potentials in each of the three conditions superimposed on
each other, on a given sensor location.
- Problem of overlapping ERPs can be seen clearly from the labels on the FCz plot.
‘Compare’ C1 plot with the one to the left:
- Initial drift between -400 and 0 evens out.
- No trace of N1 & P1 components.
- Differences emerge at ~170 ms and 350 ms
post-T2.
3
Global differences between seen and unseen trials emerge after 170 ms post-T2.
- The supplementary video gives a global overview of events specific to T2 perception (seen –
absent & unseen – absent) and of the differences between these two (seen – unseen; right map).
The authors proceed to give a detailed
overview of the sequence of events.
Source localization is used to obtain an
idea about the neural generators
underlying the observed voltage maps.
--- Treat those with benign neglect.
The difference at 170 ms is fairly
specific to one region of the
voltage map.
At this location, a positive
modulation is present for the
seen T2s, but not for the unseen
ones. “P170”
4
At around 270 ms a leftposterior negativity shows up.
The authors refer to this as the
N2 component, associated with
perceptual similarity but also
control processes (inhibition).
The authors do not emphasize
a left temporal positivity that
also looks to be unique to seen
T2s.
By 300 ms post-T2 a posterior
negativity, labeled the N3,
follows the N2.
Note how the difference is quite
sharp; the effect of visibility on
the N3 can be described as
‘all-or-nothing’.
5
Interestingly, a modulation of the
N4 component occurs around 350
ms. Given the results of Vogel et al.
this is remarkable; semantic processing of T2 should be fairly intact.
The change between seen and
unseen is gradual.
Around 430 ms the P3 waves
show a large divergence, as
expected. The P3 is associated
with target-related processing,
and memory consolidation in
particular.
The P3a (the anterior first part
of the late P3 complex) basically
appears only when T2 is well
perceived, which was also true
for the N3.
6
At 580 ms, the P3b (posterior)
is strongly present for seen T2s,
and completely absent for
unseen ones.
Are there problems with the comparisons made here?
- One might quibble that the number of observations in seen and unseen conditions was not equal.
In ERP studies, this can be a major confound as variability of the average waveforms is strongly
affected by it and may lead to the detection of spurious components.
--- The riposte is that early attentional components match perfectly (baseline level is further proof),
which makes it highly unlikely that an inequality in variance could account for the results.
The authors continue to look at T1-related activity and posit that T1-evoked potentials
interfere with T2 components. This seemed somewhat post-hoc and without a strong
contribution to the understanding of the blink and the neural processes leading up to it.
7
- I do not believe the blink arises because ‘the T1 P3b hampers the T2 N2’.
- In functional terms, the consolidation of T1 is thought to pose a bottleneck for the same process for
T2. This is already well known and somewhat beside the main point of this present paper.
Conclusions
What do these ERPs tell us?
Conscious report involves large-scale ‘late’ brain activity. In tasks like this, conscious report
cannot be due to activation in ‘early’ stimulus-specific areas.
The authors replicated (and strengthened) earlier indications that unseen events can
still evoke high-level processing (N4), although N4 amplitude was suppressed somewhat.
Support for classic two-stage models of the AB was obtained. Components associated with
stage-2 processing diverge after less than 300 ms, which strongly suggests a relation to the
observed behavioral effect (i.e., a failure to perform Task2).
Detailed sequence of events shows how P1 & N1 are preserved, N2 and N4 are linearly
modulated (reduced), and N3, P3a, & P3b eliminated for unseen T2s; providing a starting
point for further functional research tapping into these components.
8