Wednesday, May 14, 2008

Spoken Word Memory Traces within the Human Auditory Cortex

This looks like an interesting study using fMRI repetition suppression methods to identify neural networks involved in spoken word recognition. It appears such effects were found bilaterally in the superior temporal gyrus. Worth a close look!

Spoken Word Memory Traces within the Human Auditory Cortex Revealed by Repetition Priming and Functional Magnetic Resonance Imaging
Pierre Gagnepain, Gael Chetelat, Brigitte Landeau, Jacques Dayan, Francis Eustache, and Karine Lebreton
J. Neurosci. 2008;28 5281-5289


Anonymous said...

I'm not a big fan of this paper, for a couple of reasons;

1) Take a look at figure 3. They authors did not find a repetition priming effect for pseudowords, either in terms of RT or Accuracy. Behaviourally, F values for repetition effects for Pseudowords are in the region of 0.4 for both latency and accuracy differences. They don't report p for these tests, but p must be about .5

2) The authors do not report a simple "Words Primed - Words Unprimed" contrast. This is the only contrast for which they have a behavioual priming effect, and would be interpretable. They only report activation for the repetition/stimulus class interaction, but as they don't find a priming effect for pseudowords, this interaction is difficult to interpret.

3) Accuracy on the lexical decision task for words is not a great deal better than chance (about 65% on unprimed items). I think the authors only analyse correct trials, but even so, with a "yes"/"no" task with such a low accuracy rate, a high proportion of the 'correct' responses are likely to be guesses too. For pseudowords, it's much better (around 80%) suggesting that subjects were biased towards responding 'no' to stimuli during the experiment. This is probably because of the degraded stimuli and the instruction to be fast rather than accurate. Again, I think this makes it difficult to interpret the results.

4) In the absence of any a priori hypotheses, I think the statistical thresholds used are probably too lenient (and inconsistent).

Sorry, that's more than a couple of reasons. Perhaps I'm in a bad mood today. One thing I did like, unrelated to analysis of priming effects, is that they found a main effect of words vs pseudowords (increased activation for words vs pseudowords in left IFG, left superior temporal pole, left posterior MTG). This is quite a rare finding (I think) and it's interesting that this is all on the left, even at low threshold. The stimuli are degraded, but acoustics/phonology shouldn't differ between classes. I wonder what your thoughts are on this? Why not bilateral activation?

Greg Hickok said...

Thanks for the review. I haven't yet had the time to read it carefully. It looks like you have found some serious problems. I agree, it is rare to find a lexicality effect, and interesting that it is unilateral. Assuming the effect is real (and you brought up several reasons to question the study generally), the lateralization pattern may depend on what level of processing the contrast is highlighting. I would predict bilateral activation for processes up to and including the access of phonological word forms. Beyond that, it seems things become a bit more left dominant. It is possible that the word vs. p-word contrast is highlighting something more semantic-y, leading to a lateralized effect.

Methodologically, I think it is important in these kinds of studies to design and analyze your study without group averaging. The amount of cross-subject variation in speech-related STS activation is substantial when tight contrasts are employed. Who knows, failure to find right hemi activation could result from something as trivial as more cross-subject variation in the activation foci in the right hemisphere compared to left...

Anonymous said...

I agree with your point about group averaging. It seems crazy to characterise intersubject variability as just 'noise'.
Here's a nice paper
that illustrates your point (this time with a semantic categorisation task). The answer seems to be that you need more subjects (>30?) to bring out regions with more variable focii of activation. The line has to be drawn somewhere, I suppose. I'd say if you have close to 30 subjects in a RFX analysis, and you're not finding right hemisphere activiation, it's pretty safe to conclude that the process you're investigating is strongly left-lateralised.

Greg Hickok said...

Nice paper, thanks. Here's the abstract if anyone is interested:

Group Analysis and the Subject Factor in Functional
Magnetic Resonance Imaging: Analysis of Fifty Right-
Handed Healthy Subjects in a Semantic Language Task

Mohamed L. Seghier,1,2 Franc¸ ois Lazeyras,1 Alan J. Pegna,3,4,5
Jean-Marie Annoni,4 and Asaid Khateb3,4,5*

Abstract: Before considering a given fMRI paradigm as a valid clinical tool, one should first assess the reliability of functional responses across subjects by establishing a normative database and defining a
reference activation map that identifies major brain regions involved in the task at hand. However, the definition of such a reference map can be hindered by inter-individual functional variability. In this
study, we analysed functional data obtained from 50 healthy subjects during a semantic language task
to assess the influence of the number of subjects on the reference map and to characterise inter-individual functional variability. We first compared different group analysis approaches and showed that the
extent of the activated network depends not only on the choice of the analysis approach but also on
the statistical threshold used and the number of subjects included. This analysis suggested that, while
the RFX analysis is suitable to detect confidently true positive activations, the other group approaches are useful for exploratory investigations in small samples. The application of quantitative measures at the voxel and regional levels suggested that while 15–20 subjects were sufficient to reveal reliable and robust left hemisphere activations, >30 subjects were necessary for revealing more variable and weak right hemisphere ones. Finally, to visualise inter-individual variability, we combined two similarity indices that assess the percentages of true positive and false negative voxels in individual activation patterns relative to the group map. We suggest that these measures can be used for the estimation of
the degree of ‘normality’ of functional responses in brain-damaged patients, where this question is often raised, and recommend the use of different quantifications to appreciate accurately the inter-individual functional variability that can be incorporated in group maps. Hum Brain Mapp 29:461–477, 2008.