Saaussan -
This is an important question that I am not qualified to
answer fully, but a couple of related issues are:
1. If you have globally scaled, and your covariates are
correlated with the global estimate (eg if many brain
areas were activated versus a low-level baseline, the
global estimate can become contaminated by local changes),
the deactivations can only really be interpreted relative
to the global signal (see Andersson, 1997; Aguirre et al
1998).
2. Even without global normalisation, there are some
experimental designs where you do not estimate the
event-related response efficiently. If you have short
SOA with no null events for example, you can be sensitive
to differences between event-types, as in Darren's
example, but not the sum (mean) effect versus baseline,
and therefore get several patterns that maintain a
relative activation of event-type1 to event-type2, but
vary in the activation/deactivations versus baseline
(more generally, when the covariates of interest are
highly correlated). Thus you should always check that
the deactivations are themselves significant (eg [-1 0]
and [0 -1] contrasts).
3. Even without the above two problems, the interpretation
of event-related deactivations is, I believe, still a
matter of debate. I think a transient reduction in
baseline neural activity upon stimulus presentation
could produce an inverted BOLD response. Fransson et
al (1999), Neuroimage, 8, 611-618, might be a good
reference to start with.
Hope this helps
Rik
> >We know that the BOLD signal reflects hemodynamic changes as a result of
> >neural activity changes.
> >My questions are
> >-Does positive activation necessarily mean increased stimulation (more
> >firing)?
> >By the same token, deactivation means decreased stimulation, or possibly
> >increased inhibition.
> >
> >-Are there any references that i can look , to find an answer for that.
> >
>
> ______
>
> It is something we
> have discussed and struggled with among our group at Northwestern
> quite a bit.
>
> For example sometimes you see curves that look like this
>
> *** *** x
> x * * xx
> x * * * x
> x x
> x x
> xx x
> x
>
> The point is that the asterisks and + curves are clearly separate
> (and for arguments sake we'll say significantly different. Thus the
> contrast 1 (*) vs.
> -1 (x) would show "activation" of the areas corresponding to the *
> task. BUT as you can see both curves go down meaning a decrease in
> BOLD signal.
>
> Physiologically does this decrease mean a reduction in neuronal
> firing? Increased inhibition by the way might look at an activation
> of the inhibiting area and then presumably reduced activation at the
> site affected by the inhibition.
>
> BOLD signal reflects change in local deoxyhemoglobin concentrations.
> Are there other influences on this signal? Is it possible that blood
> flow becomes uncoupled from neuronal firing even in the normal
> individual during certain types of tasks? For example given that some
> amount of vascular perfusion to an area appears to be a luxury what
> if neuronal firing increased but other influences on local
> hemodynamics such as acetylcholine decreased- would we see a decline
> in signal despite the fact that neuronal firing had increased
> somewhat.
>
> Anyway- a very interesting question. I would encourage you to post
> this on the SPM list and you can include my response, such as it is,
> if you want. If you do gain more insight into this I'd like to know.
>
> Darren
--
---------------------------8-{)}-------------------------
DR R HENSON EMAIL [log in to unmask]
Wellcome Department of
Cognitive Neurology TEL (work1) +44 (0)20 7833 7483
12 Queen Square TEL (work2) +44 (0)20 7833 7472
London, WC1N 3BG FAX +44 (0)20 7813 1420
URL: http://www.psychol.ucl.ac.uk/rik.henson/index.html
---------------------------------------------------------
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|