- Email: [email protected]
Essential ingredients of imaging
TICS Aug 2000 _
Update
14/7/00
10:10 am
Page 296
Kourtzi and Kanwisher – Response to David and Senior
Comment
Talairach coordinates is misleading because this coordinate system is based on the necessarily imperfect alignment of physically different brains, and because it specifies points in the brain whereas activations generally cover large and irregularly shaped regions. Third, David and Senior comment that ‘the lack of a statistical test of coincident activation between the “pure” visual motion versus the implied motion is unfortunate’. If they are referring to a test of the coincidence of motion selectivity and implied-motion selectivity, we accomplished this goal with our finding of a significantly higher response to static images with implied motion than static images without implied motion in areas that respond significantly to stimulus motion. If they are instead referring to a direct statistical comparison of the response to stimulus motion versus that to static images with implied motion, it is not clear what important theoretical question such a test would answer. Fourth, David and Senior comment that ‘the use of such a potent stimulus to induce activity in MT/MST may have maximized the chances of observing overlapping activations’. Maximizing the chance of detecting what you are looking for is generally considered desirable in experimental design provided you minimize confounds. Our test of the response to implied motion did not indiscriminately activate visual areas and David and Senior raise no clear objection to it. Moreover, the MT localizer used in our studies (low contrast moving versus stationary rings) activates MT more selectively than other, higher-contrast stimuli that might also activate regions beyond MT involved in motion processing8. Thus, we reject all of the putative pitfalls in our study that David and Senior raise. However, we are in turn
concerned about several aspects of the Senior et al. study6. First, the reliance on a group analysis in the Senior et al. study results in substantial blurring of the functional images; this blurring can cause a spurious overlap in brain activations. We avoided this problem by using analyses of regions of interest defined individually for each subject. Second, Senior et al. did not attempt to control for attentional confounds that might be expected to result from the fact that the images with implied motion are more interesting than the stills without implied motion. Indeed, the activations observed in the regions posterior to MT1 in the Senior et al. study might be better explained by such attentional confounds than in terms of the semantic processing of implied-motion images postulated by David and Senior2. Our study minimized these problems by including a task (repetition detection) that requires attention to all stimuli. Third, David and Senior argue that the 400–600 ms reaction times found in their behavioral test of implied motion are ‘rather fast to allow for much cognitive elaboration’, and therefore activation of MT is more likely to reflect processes that occur within the visual system itself (rather than as a result of feedback from higher areas). However, research in cognitive psychology has provided ample evidence of high-level cognitive processes that are unlikely to be computed within the visual system but that can nonetheless be carried out in less than half a second (e.g. Refs 9,10). So the fast reaction times in the behavioral implied-motion task do not speak to the question of whether it is MT1 itself, or a ‘higher’ area, that extracts the information about implied motion. An answer to this question will require the use of a technique with higher temporal resolution than fMRI.
It is encouraging that despite the methodological differences in the two studies, largely similar results were obtained. These findings converge on the important conclusion that high-level perceptual inferences can substantially influence activity in extrastriate cortex.
References 1 Kourtzi,
Z.
and
Kanwisher,
N.
(2000)
Activation of human MT/MST by static images with implied motion. J. Cogn. Neurosci. 12, 48–55 2 David, A.S. and Senior, C. (2000) Implicit motion and the brain. Trends Cognit. Sci. 4, 293–295 3 Tootell, R.B.H. et al. (1995) Functional analysis of human MT and related visual cortical areas using magnetic resonance imaging. J. Neurosci. 15, 3215–3230 4 Watson, J.D.G. et al. (1993) Area V5 of the human brain: evidence from a combined study using positron emission tomography and magnetic resonance imaging. Cereb. Cortex 3, 79–94 5 Zeki, S. et al. (1991) A direct discrimination of functional specialization in human visual cortex. J. Neurosci. 11, 641–649 6 Senior, C. et al. (2000) The functional neuroanatomy of implicit-motion perception or ‘representational momentum’. Curr. Biol.10, 16–22 7 Allison T. et al. (2000) Social perception from visual cues: role of the STS region. Trends Cognit. Sci. 4, 265–277 8 Tootell, R.B.H. et al. (1997) Functional analysis of V3A and related areas in human visual cortex. J. Neurosci. 15, 7060–7078 9 Kutas, M. and Hillyard, S.A. (1980) Reading senseless sentences: brain potentials reflect semantic incongruity. Science 207, 203–205 10 Rayner, K., ed. (1983) Eye Movements in Reading : Perceptual and Language Processes, Academic Press
Meetings
Essential ingredients of imaging The fMRI Experience II , 11–12 May 2000, Weston Education Centre, Kings College Hospital, Denmark Hill, London, UK.
F
unctional magnetic resonance imaging (fMRI) is rapidly becoming the technique of choice for many investigators in the clinical and cognitive neurosciences. The growing demand for this technology spurred the initiative to organize a meeting that would bring together some of the foremost practitioners of fMRI to share their knowledge and experience to a wide audience of students and established researchers. The timing of this meeting at the outset of a new millennium was also an opportune moment to review the contribution of fMRI to our understanding of human brain function and
296
to speculate on its potential capabilities in the future. The ‘fMRI Experience II’ was the first forum to bring together experts in functional neuroimaging from four centres of excellence in the UK; the Institute of Psychiatry (IOP), King’s College London; the Functional Imaging Laboratory (FIL) at the Wellcome Department of Cognitive Neurology, University College London; the Centre for Functional Imaging of the Brain (FMRIB) at the University of Oxford; and the Brain Imaging Unit at the University of Cambridge. The meeting evolved from two smaller workshops
1364-6613/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. Trends in Cognitive Sciences – Vol. 4, No. 8,
held at Oxford and the IOP last year. The first day consisted of teaching sessions in which experts described basic principles and state-of-the-art developments in fMRI. On the second day, graduate students gave oral presentations on a diverse range of topics from innovative neurocognitive studies to sophisticated advances in signal processing and data-analytic techniques. This was a valuable component of the meeting as it afforded a rare opportunity for PhD students to make oral presentations and to obtain helpful feedback on their research from an audience of peers and imaging specialists.
PII: S1364-6613(00)01514-X
August 2000
TICS Aug 2000 _
14/7/00
10:10 am
Page 297
Fu et al. – Essential ingredients of imaging
Update Meetings
Despite the rain that welcomed our registrants, Peter Jezzard (FMRIB, London) brightly opened the meeting with a comprehensive introduction to magnetic resonance physics and a discussion of the underlying neurophysiology that contributes to the fMRI signal in the form of a blood oxygenation level dependent (BOLD) response. This was followed by two presentations describing how MR signals are processed to produce statistical maps of brain activity. Karl Friston (FIL, London) gave an overview of the principles underpinning statistical parametric mapping (SPM). The widespread use of SPM reflects the fact that it was the first data-analysis package to include a satisfactory solution to the multiplecomparison problem inherent in voxelwise image analysis. This solution uses Gaussian field theory to model the effects of spatial correlations on activation maps, thereby avoiding the unnecessarily harsh Bonferroni correction. Mick Brammer (IOP, London) discussed another method of inference that is also commonly used, that is, non-parametric approaches based on random permutation of data. This method has long been acknowledged to be free from many of the assumptions of parametric statistical techniques, but its widespread use has been limited until recently by its heavy computational demands. Today’s cheap, high-speed computing should now, Brammer argued, lead to more extensive use of these techniques in functional imaging. Although data analysis is the backbone of functional neuroimaging, this meeting was the first general forum which brought together and facilitated an open discussion of the advantages and disadvantages of various methods. The morning session concluded with a series of edifying presentations by Gemma Calvert (FMRIB, Oxford), Chris Andrew and Vincent Giampietro (both from the IOP, London). Gemma Calvert detailed the multiple steps involved in the design and analysis of an fMRI study, highlighting the traps and pitfalls that can be encountered at each stage and providing suggestions as to how potential problems might be avoided. Chris Andrew followed with an interactive demonstration of some of the latest fMRI paradigms and emphasized the importance of considering the subject’s perspective when designing fMRI experiments. Finally, using the analogy of a cooking lesson, Vincent Giampietro completed the morning by providing a lucid description of the many stages involved in fMRI data analysis that often confound newcomers to the field. The afternoon session opened with a presentation by Cathy Price (FIL, London) on some of the more sophisticated fMRI paradigms in current use. With examples from her own research on the neural basis of language processing, Price described the implications of using categorical subtractions, conjunctions, parametric variables and
interactions, and their effects on the interpretation of activation maps. One of the most recent and exciting developments in fMRI design has been the introduction of event-related paradigms. Most early fMRI studies used a ‘block’ design, in which an ‘experimental’ condition is presented in alternating blocks with a ‘baseline’ condition. The advantage of this design is the robustness and power of the BOLD responses that are elicited. However, responding in ‘blocks’ is not how we typically behave in day-to-day activities. Rik Henson (FIL, London) and Edson Amaro Jr (Department of Radiology, University of São Paulo, Brazil, and IOP, London) made tandem presentations on eventrelated fMRI (erfMRI). This method is more akin to psychophysical paradigms in which trials are presented and analysed independently; this permits different trial types to be interleaved in a random sequence. Henson described strategies for optimizing experimental design in erfMRI, and Amaro discussed how BOLD responses could be deconvolved and analysed – this is necessary because their interactions rapidly become quite complicated as the rate of stimulus presentation is increased. Although research using fMRI has grown exponentially since its inception less than a decade ago, it nevertheless has certain limitations as a neuroimaging technique. A key disadvantage of fMRI relative to other methods, such as positron emission tomography (PET) and single photon emission tomography (SPET), is that image acquisition is associated with high-amplitude acoustic noise (about 110 dB); this makes it difficult for subjects to hear auditory stimuli or their own voice during scanning. Philip McGuire (IOP, London) illustrated various acquisition sequences that minimize the confound of scanner noise with examples from recent auditory–verbal paradigms, including the use of discontinuous acquisition sequences that introduce periods of temporary silence. Ed Bullmore (Cambridge, UK and IOP, London) followed this by describing the potential application of fMRI to the investigation of psychopharmacological drug action. Bullmore cautioned that although fMRI offers advantages of sensitivity, repeatability and catholicism of experimental design compared to PET and SPET there are potentially confounding drug effects on cerebral blood vessels. Steve Williams (IOP) also discussed some of the limitations with existing MRI hardware and software in his talk on advanced physics. Moreover, Williams focused on the exciting future developments and clinical applications of ‘hyphenated’ techniques that include combined PET and fMRI, diffusion tensor imaging (DTI) and fMRI, and EEG and fMRI. The penultimate speaker for the first day was Derek Jones (IOP, London and University of Leicester, UK), who described the corollary functional
neuroimaging technique of DTI. This technique can discriminate the orientation and integrity of white matter tracts that are not routinely discernable with conventional MR imaging and is thus a useful complement to the nodal, regional activation data obtained with fMRI. Jones’ presentation will be remembered for the erudite manner in which he was able to demonstrate this highly complex technique by an analogy to the principles involved in preparing spaghetti pomodoro. The day was concluded with an overview and predictions about the future of fMRI research by Karl Friston and Steve Williams. On the second day, students gave oral presentations on their recent research. There were 24 presentations divided into the following sessions: methodology, primary sensory systems, sensorimotor function, emotion, and cognition in health and disorders. The emphasis of the day was to encourage and assist new researchers to make oral presentations of their work to a discerning yet friendly academic audience. The student presenters came from throughout the UK, the USA (South Carolina), Leuven, Belgium and Zurich, Switzerland. Both the calibre of research and the quality of presentations was impressive. [Abstracts for all the presentations from both days can be found at our website: www.iop. kcl.ac.uk/IoP/Teaching/fMRI/info.stm]. The fMRI Experience II brought together over 300 registrants and speakers from across London; the UK, Belgium, Bulgaria, Denmark, France, Italy, The Netherlands, Slovakia, Switzerland, and even the USA and Brazil. The interest and support for this meeting was outstanding. Significant credit must be given to the speakers and chairs for volunteering their time and to the student presenters for their excellent research. The waiving of registration fees was made possible by generous financial support from the Wellcome Trust, which also allowed lunch to be provided for all the delegates. From the enthusiasm of the expert speakers and session chairs and the quality of the students’ new research presentations, it is clear that fMRI research is burgeoning, even though it is still much in its infancy. We look forward with anticipation to ‘The fMRI Experience III’. Cynthia H.Y. Fu, Carl Senior, Tamara Russell Institute of Psychiatry, King’s College London, de Crespigny Park, London, UK SE5 8AF; tel: 144 20 7848 0514 fax: 144 20 7701 9044 e-mail: [email protected] Sarah-Jayne Blakemore Wellcome Department of Cognitive Neurology, Institute of Neurology, University College London, 12 Queen Square, London UK WC1N 3BG; Gemma Calvert Centre for Functional Magnetic Resonance Imaging of the Brain (FMRIB), University of Oxford, John Radcliffe Hospital, Oxford, UK OX3 9DU.
297 Trends in Cognitive Sciences – Vol. 4, No. 8,
August 2000
Update
14/7/00
10:10 am
Page 296
Kourtzi and Kanwisher – Response to David and Senior
Comment
Talairach coordinates is misleading because this coordinate system is based on the necessarily imperfect alignment of physically different brains, and because it specifies points in the brain whereas activations generally cover large and irregularly shaped regions. Third, David and Senior comment that ‘the lack of a statistical test of coincident activation between the “pure” visual motion versus the implied motion is unfortunate’. If they are referring to a test of the coincidence of motion selectivity and implied-motion selectivity, we accomplished this goal with our finding of a significantly higher response to static images with implied motion than static images without implied motion in areas that respond significantly to stimulus motion. If they are instead referring to a direct statistical comparison of the response to stimulus motion versus that to static images with implied motion, it is not clear what important theoretical question such a test would answer. Fourth, David and Senior comment that ‘the use of such a potent stimulus to induce activity in MT/MST may have maximized the chances of observing overlapping activations’. Maximizing the chance of detecting what you are looking for is generally considered desirable in experimental design provided you minimize confounds. Our test of the response to implied motion did not indiscriminately activate visual areas and David and Senior raise no clear objection to it. Moreover, the MT localizer used in our studies (low contrast moving versus stationary rings) activates MT more selectively than other, higher-contrast stimuli that might also activate regions beyond MT involved in motion processing8. Thus, we reject all of the putative pitfalls in our study that David and Senior raise. However, we are in turn
concerned about several aspects of the Senior et al. study6. First, the reliance on a group analysis in the Senior et al. study results in substantial blurring of the functional images; this blurring can cause a spurious overlap in brain activations. We avoided this problem by using analyses of regions of interest defined individually for each subject. Second, Senior et al. did not attempt to control for attentional confounds that might be expected to result from the fact that the images with implied motion are more interesting than the stills without implied motion. Indeed, the activations observed in the regions posterior to MT1 in the Senior et al. study might be better explained by such attentional confounds than in terms of the semantic processing of implied-motion images postulated by David and Senior2. Our study minimized these problems by including a task (repetition detection) that requires attention to all stimuli. Third, David and Senior argue that the 400–600 ms reaction times found in their behavioral test of implied motion are ‘rather fast to allow for much cognitive elaboration’, and therefore activation of MT is more likely to reflect processes that occur within the visual system itself (rather than as a result of feedback from higher areas). However, research in cognitive psychology has provided ample evidence of high-level cognitive processes that are unlikely to be computed within the visual system but that can nonetheless be carried out in less than half a second (e.g. Refs 9,10). So the fast reaction times in the behavioral implied-motion task do not speak to the question of whether it is MT1 itself, or a ‘higher’ area, that extracts the information about implied motion. An answer to this question will require the use of a technique with higher temporal resolution than fMRI.
It is encouraging that despite the methodological differences in the two studies, largely similar results were obtained. These findings converge on the important conclusion that high-level perceptual inferences can substantially influence activity in extrastriate cortex.
References 1 Kourtzi,
Z.
and
Kanwisher,
N.
(2000)
Activation of human MT/MST by static images with implied motion. J. Cogn. Neurosci. 12, 48–55 2 David, A.S. and Senior, C. (2000) Implicit motion and the brain. Trends Cognit. Sci. 4, 293–295 3 Tootell, R.B.H. et al. (1995) Functional analysis of human MT and related visual cortical areas using magnetic resonance imaging. J. Neurosci. 15, 3215–3230 4 Watson, J.D.G. et al. (1993) Area V5 of the human brain: evidence from a combined study using positron emission tomography and magnetic resonance imaging. Cereb. Cortex 3, 79–94 5 Zeki, S. et al. (1991) A direct discrimination of functional specialization in human visual cortex. J. Neurosci. 11, 641–649 6 Senior, C. et al. (2000) The functional neuroanatomy of implicit-motion perception or ‘representational momentum’. Curr. Biol.10, 16–22 7 Allison T. et al. (2000) Social perception from visual cues: role of the STS region. Trends Cognit. Sci. 4, 265–277 8 Tootell, R.B.H. et al. (1997) Functional analysis of V3A and related areas in human visual cortex. J. Neurosci. 15, 7060–7078 9 Kutas, M. and Hillyard, S.A. (1980) Reading senseless sentences: brain potentials reflect semantic incongruity. Science 207, 203–205 10 Rayner, K., ed. (1983) Eye Movements in Reading : Perceptual and Language Processes, Academic Press
Meetings
Essential ingredients of imaging The fMRI Experience II , 11–12 May 2000, Weston Education Centre, Kings College Hospital, Denmark Hill, London, UK.
F
unctional magnetic resonance imaging (fMRI) is rapidly becoming the technique of choice for many investigators in the clinical and cognitive neurosciences. The growing demand for this technology spurred the initiative to organize a meeting that would bring together some of the foremost practitioners of fMRI to share their knowledge and experience to a wide audience of students and established researchers. The timing of this meeting at the outset of a new millennium was also an opportune moment to review the contribution of fMRI to our understanding of human brain function and
296
to speculate on its potential capabilities in the future. The ‘fMRI Experience II’ was the first forum to bring together experts in functional neuroimaging from four centres of excellence in the UK; the Institute of Psychiatry (IOP), King’s College London; the Functional Imaging Laboratory (FIL) at the Wellcome Department of Cognitive Neurology, University College London; the Centre for Functional Imaging of the Brain (FMRIB) at the University of Oxford; and the Brain Imaging Unit at the University of Cambridge. The meeting evolved from two smaller workshops
1364-6613/00/$ – see front matter © 2000 Elsevier Science Ltd. All rights reserved. Trends in Cognitive Sciences – Vol. 4, No. 8,
held at Oxford and the IOP last year. The first day consisted of teaching sessions in which experts described basic principles and state-of-the-art developments in fMRI. On the second day, graduate students gave oral presentations on a diverse range of topics from innovative neurocognitive studies to sophisticated advances in signal processing and data-analytic techniques. This was a valuable component of the meeting as it afforded a rare opportunity for PhD students to make oral presentations and to obtain helpful feedback on their research from an audience of peers and imaging specialists.
PII: S1364-6613(00)01514-X
August 2000
TICS Aug 2000 _
14/7/00
10:10 am
Page 297
Fu et al. – Essential ingredients of imaging
Update Meetings
Despite the rain that welcomed our registrants, Peter Jezzard (FMRIB, London) brightly opened the meeting with a comprehensive introduction to magnetic resonance physics and a discussion of the underlying neurophysiology that contributes to the fMRI signal in the form of a blood oxygenation level dependent (BOLD) response. This was followed by two presentations describing how MR signals are processed to produce statistical maps of brain activity. Karl Friston (FIL, London) gave an overview of the principles underpinning statistical parametric mapping (SPM). The widespread use of SPM reflects the fact that it was the first data-analysis package to include a satisfactory solution to the multiplecomparison problem inherent in voxelwise image analysis. This solution uses Gaussian field theory to model the effects of spatial correlations on activation maps, thereby avoiding the unnecessarily harsh Bonferroni correction. Mick Brammer (IOP, London) discussed another method of inference that is also commonly used, that is, non-parametric approaches based on random permutation of data. This method has long been acknowledged to be free from many of the assumptions of parametric statistical techniques, but its widespread use has been limited until recently by its heavy computational demands. Today’s cheap, high-speed computing should now, Brammer argued, lead to more extensive use of these techniques in functional imaging. Although data analysis is the backbone of functional neuroimaging, this meeting was the first general forum which brought together and facilitated an open discussion of the advantages and disadvantages of various methods. The morning session concluded with a series of edifying presentations by Gemma Calvert (FMRIB, Oxford), Chris Andrew and Vincent Giampietro (both from the IOP, London). Gemma Calvert detailed the multiple steps involved in the design and analysis of an fMRI study, highlighting the traps and pitfalls that can be encountered at each stage and providing suggestions as to how potential problems might be avoided. Chris Andrew followed with an interactive demonstration of some of the latest fMRI paradigms and emphasized the importance of considering the subject’s perspective when designing fMRI experiments. Finally, using the analogy of a cooking lesson, Vincent Giampietro completed the morning by providing a lucid description of the many stages involved in fMRI data analysis that often confound newcomers to the field. The afternoon session opened with a presentation by Cathy Price (FIL, London) on some of the more sophisticated fMRI paradigms in current use. With examples from her own research on the neural basis of language processing, Price described the implications of using categorical subtractions, conjunctions, parametric variables and
interactions, and their effects on the interpretation of activation maps. One of the most recent and exciting developments in fMRI design has been the introduction of event-related paradigms. Most early fMRI studies used a ‘block’ design, in which an ‘experimental’ condition is presented in alternating blocks with a ‘baseline’ condition. The advantage of this design is the robustness and power of the BOLD responses that are elicited. However, responding in ‘blocks’ is not how we typically behave in day-to-day activities. Rik Henson (FIL, London) and Edson Amaro Jr (Department of Radiology, University of São Paulo, Brazil, and IOP, London) made tandem presentations on eventrelated fMRI (erfMRI). This method is more akin to psychophysical paradigms in which trials are presented and analysed independently; this permits different trial types to be interleaved in a random sequence. Henson described strategies for optimizing experimental design in erfMRI, and Amaro discussed how BOLD responses could be deconvolved and analysed – this is necessary because their interactions rapidly become quite complicated as the rate of stimulus presentation is increased. Although research using fMRI has grown exponentially since its inception less than a decade ago, it nevertheless has certain limitations as a neuroimaging technique. A key disadvantage of fMRI relative to other methods, such as positron emission tomography (PET) and single photon emission tomography (SPET), is that image acquisition is associated with high-amplitude acoustic noise (about 110 dB); this makes it difficult for subjects to hear auditory stimuli or their own voice during scanning. Philip McGuire (IOP, London) illustrated various acquisition sequences that minimize the confound of scanner noise with examples from recent auditory–verbal paradigms, including the use of discontinuous acquisition sequences that introduce periods of temporary silence. Ed Bullmore (Cambridge, UK and IOP, London) followed this by describing the potential application of fMRI to the investigation of psychopharmacological drug action. Bullmore cautioned that although fMRI offers advantages of sensitivity, repeatability and catholicism of experimental design compared to PET and SPET there are potentially confounding drug effects on cerebral blood vessels. Steve Williams (IOP) also discussed some of the limitations with existing MRI hardware and software in his talk on advanced physics. Moreover, Williams focused on the exciting future developments and clinical applications of ‘hyphenated’ techniques that include combined PET and fMRI, diffusion tensor imaging (DTI) and fMRI, and EEG and fMRI. The penultimate speaker for the first day was Derek Jones (IOP, London and University of Leicester, UK), who described the corollary functional
neuroimaging technique of DTI. This technique can discriminate the orientation and integrity of white matter tracts that are not routinely discernable with conventional MR imaging and is thus a useful complement to the nodal, regional activation data obtained with fMRI. Jones’ presentation will be remembered for the erudite manner in which he was able to demonstrate this highly complex technique by an analogy to the principles involved in preparing spaghetti pomodoro. The day was concluded with an overview and predictions about the future of fMRI research by Karl Friston and Steve Williams. On the second day, students gave oral presentations on their recent research. There were 24 presentations divided into the following sessions: methodology, primary sensory systems, sensorimotor function, emotion, and cognition in health and disorders. The emphasis of the day was to encourage and assist new researchers to make oral presentations of their work to a discerning yet friendly academic audience. The student presenters came from throughout the UK, the USA (South Carolina), Leuven, Belgium and Zurich, Switzerland. Both the calibre of research and the quality of presentations was impressive. [Abstracts for all the presentations from both days can be found at our website: www.iop. kcl.ac.uk/IoP/Teaching/fMRI/info.stm]. The fMRI Experience II brought together over 300 registrants and speakers from across London; the UK, Belgium, Bulgaria, Denmark, France, Italy, The Netherlands, Slovakia, Switzerland, and even the USA and Brazil. The interest and support for this meeting was outstanding. Significant credit must be given to the speakers and chairs for volunteering their time and to the student presenters for their excellent research. The waiving of registration fees was made possible by generous financial support from the Wellcome Trust, which also allowed lunch to be provided for all the delegates. From the enthusiasm of the expert speakers and session chairs and the quality of the students’ new research presentations, it is clear that fMRI research is burgeoning, even though it is still much in its infancy. We look forward with anticipation to ‘The fMRI Experience III’. Cynthia H.Y. Fu, Carl Senior, Tamara Russell Institute of Psychiatry, King’s College London, de Crespigny Park, London, UK SE5 8AF; tel: 144 20 7848 0514 fax: 144 20 7701 9044 e-mail: [email protected] Sarah-Jayne Blakemore Wellcome Department of Cognitive Neurology, Institute of Neurology, University College London, 12 Queen Square, London UK WC1N 3BG; Gemma Calvert Centre for Functional Magnetic Resonance Imaging of the Brain (FMRIB), University of Oxford, John Radcliffe Hospital, Oxford, UK OX3 9DU.
297 Trends in Cognitive Sciences – Vol. 4, No. 8,
August 2000