I am a PhD student and am about to embark on my own set of fMRI experiments into consciousness and attention. I have heard Max speak a number of times on the issue and should direct your attention to the replies by Rik Henson (they can be downloaded from his webpage).

Max however has clearly been selective in the examples he has provided. One of the great success stories of Cognitive Neuroscience has been understanding the neural processes involved in vision and attention. Pure cognitive models of vision have been built implicitly on the notion that vision is only ‘feedforward’ in that simple processes feedforward into more complicted ones and then eventually an attentional filter is met which only allows a restricted amount of information into consciousness.
Neurophysiologists and imagers over the past decade however have been wonderfully successful in largely dispelling that view held by old guard vision scientists. What we know now is that vision is massively interactive, with higher level processes feeding information back into lower level processes. In addition we now know that the correlate of attention is the feedback of information from decision related and spatial related areas of the brain into the visual areas. This massively interactive view of attention and perception has been carried along largely by physiologists and neuroimagers.
The future for cognitive neuroscientists is neurodistruption. Chambers and Mattingly (2005) suggest that neuroimaging is a liberal procedure, often detecting activation in regions of the brain that are not actually involved in the task under investigation. Chambers and Mattingly argue that imaging should be followed up by neurodistruption. Neurodistruption involves using Transcranial Magnetic Stimulation to excite populations of neurons in a very small region of the cortex. This wholesale excitation of this population renders the area computationally useless (similar to adding noise to an image). Eg., applying TMS to Broca’s area arrests speech (albetit very briefly). So by using TMS we can infer whether a reigion really is involved in a certain task. Best yet, TMS only disrupts neural activity for about 10ms, so the feedforward and feeback relationships can be determined (especially if you have two TMS coils!) You can see Pascal-Leone 2001 Science experiment showing the feedforward and feedback relationship between two visual areas.
Cognitive Neuroscience is only new, and already it has provided a wealth of evidence about how vision works.

MCCABE, D., CASTEL, A. (2008). Seeing is believing: The effect of brain images on judgments of scientific reasoning. Cognition, 107(1), 343-352

Basically they have found that when people read a science article, they rate the same (flawed) stories to be more scientifically sound when they are accompanied by a brain image, compared to a bar chart. The authors conclude that people have a "natural affinity for reductionistic explanations of cognitive phenomena".

This may also mean that it is easier to get money for this type of studies. If so, this also goes some way to explain some resentment from other scientists. In a world of finite resources, funding for an fMRI project may mean no money for tens of other projects (given how expensive fMRI can be).

cheers

A few other places where I think neuroimaging can help understand things that previous psych work has not:
-How do implicit and explicit attitude interact to result in our ultimate evaluations/ attitudes/ behaviors? See Cunningham & Zelazo (2007) for an overview of one theory that integrates past work in psych, cognitive science, and insights provided from neuroimaging.
-On a related point, how do people control race bias? See work by Phelps and colleagues.
-Distinctions between internal and external focus: see review article by Lieberman (2007) on core processes in social cognitive neuroscience.

Claim are often overstated and the work has been too exploratory for too long. Unfortunately, trying to understand an engine while it's working is darn near impossible.

It's hard for me to foresee the future of this work until the technology proceeds much farther. However, precise measurement may simply highlight another problem - no two brains work exactly alike.

Cheers,
Mike

I'm currently reading Dan Goleman's "Social Intelligence", which as far as I've read is trying to make the information we get from imaging useful, maybe you've already read it, if not it may be worth checking out.

Other thing I thought of is this video of a TED speech by V.S. Ramachandran that also is trying to make the connection between physiology and psychology. Here's the web address: http://www.spring.org.uk/2007/10/ramachandran -on-capgras-syndrome.php