By Elizabeth DuPre and Kirstie Whitaker
This month we continued our Open Science Demo Call series by speaking to Anisha Keshavan, Yaroslav O. Halchenko, and Athina Tzovara about three tools they’re developing to improve openness and access in neuroimaging research.
Anisha introduced braindr, a project she’s developed to crowdsource quality control of large datasets such as the Healthy Brain Network data set. It builds off her previous work in creating MindControl but provides a fun, Tinder-inspired interface for image ratings. She encourages anyone interested to check out the app, remix it for their own data, or contribute to the conversation on how to do quality control of images!
Yaroslav told us about DataLad, a solution devised to allow for versioning data. We’ve already recognized the importance of versioning code, but it applies to data too! As Yaroslav pointed out, data can change or have “bugs” like the dreaded left-right orientation flip in MRI data, so understanding what version you’re working with is important. Using DataLad, Yaroslav demonstrated how to install datasets from sources like OpenNeuro and discussed how it can even be used for data sets before they are made publicly available. Interested contributors are welcome to check out the code!
Athina introduced a survey she’s actively developing to better understand how research treats underrepresented minorities. It aims to allow non-scientists --- particularly those belonging to traditionally underrepresented minorities --- to take an active role in the scientific process, bridging the divide between researchers and participants. Originally developed through the Mozilla Open Leadership program, the survey is still open to feedback from the community, and Athina encourages anyone interested to join the discussion on GitHub!
Our next call will be on Thursday March 22nd at 7pm GMT (check your local time zone). If you’d like to nominate yourself or someone else to be featured on these monthly calls, please add their information at this github issue, or email the host of the calls Kirstie Whitaker at email@example.com. You can also join the OSSIG google group to receive reminders each month.
"A brain scan may reveal the neural signs of anxiety, but a Kokoschka painting, or a Schiele self-portrait, reveals what an anxiety state really feels like. Both perspectives are necessary if we are to fully grasp the nature of the mind, yet they are rarely brought together".
-- Eric Kandel
Visual art can provide a glimpse into people’s consciousness. It works as a bridge, not only connecting us to each other, but also with the past, present, and future. The act of creating art is also therapeutic, and represents a powerful resource for mental and physical well-being. Yet, the mechanisms underlying the brain’s capacity to generate art remains largely elusive. While it has been commonly reported that the right brain (posterior parietal and posterior temporal) is dominant for artistic ability, emerging literature strongly indicates that the left brain is not a silent partner. Instead, it contributes to more of the symbolic/conceptual aspects of art. Moreover, the emergence of visual artistic skills in the healthy brain has been linked to plasticity in areas (in both hemispheres) responsible for cognitive processes. Which begs the question: how is visual artistic creativity affected by neurodegeneration?
In fact, art in the context of neurodegenerative diseases (e.g. Alzheimer’s disease, frontotemporal dementia) provides a unique window into brain anatomy and function. In this interview, I discuss the link between neurodegeneration and art with Bruce Miller, director of the Memory and Aging Centre at the University of California. Bruce also oversees the unique Hellman Visiting Artist Program, created to foster dialogue between scientists, caregivers, patients, clinicians and the public regarding creativity and the brain.
Q&A WITH BRUCE MILLER
AmanPreet Badhwar (AB): Can you begin by saying something about your background?
Bruce Miller (BM): I am a behavioural neurologist at the University of California, San Francisco. I focus a lot on degenerative disease: the clinical presentation, differential diagnosis of dementia, also deep dive into frontotemporal dementia. I think a lot about behavioral phenomena, particularly early in the course of these diseases.
I started realizing the importance of art and dementia very serendipitously. It was based on seeing a single patient (Jack). The son told me his father has become an artist in the setting of the illness. And I said “of course as the disease has progressed his work has gotten worse”, and he said “oh no it has gotten better”. So he sent me a series of pictures, and I was fascinated and really enchanted by the work that he did, and began to look in detail into the visual artistic process in that patient. Jack was preoccupied with creating purple and yellow art pieces, and a phrase I often heard from him was “ yellow and purple wave over me”.
I did not think it was a coincidence, although many people around me thought it was, and I was stubborn enough to pursue this, and continued to look for it in my frontotemporal dementia and progressive aphasia population. It does not take much time to hear about somebody, who they are, what they do etc. I would argue that this should be a mandatory part of any evaluation.
AB: How do the worlds of neuroscience and art combine?
BM: Art is unique to the human species. Other animals don’t spontaneously produce art and even our predecessors like the neanderthals and homo erectus made art. There are records of very sophisticated and complex cave paintings by homo sapiens that showed animals, had three-dimensional components and colours. So we developed this ability spontaneously, and without much formal teaching. The sense is that there is something really unique that happened, there was a change in the human brain, maybe a change in human circumstances that lead to this flourishing of art, and this continues to be a part of our ancient and modern societies.
Also looking at the human output around art: some people are extraordinary, and some never produce art. So I think art is a very interesting aspect of humanity and a very interesting aspect of the human brain, and that the two things cannot be more connected.
AB: You previously stated that “creativity is one characteristic that has been observed to improve with time, both in healthy older adults and people with age-related neurodegenerative disease”. Is the trajectory for artistic creativity different in normal aging and in age-related dementias?
BM: I think it’s a very interesting, complex question, tackling aging of humans and art. We are very interested in elder artists, there is no doubt about it. Picasso was in his eighties, he produced very different but interesting pieces, but they delighted people. There is no doubt that his work was exciting. Was it better when he was young, or was it more innovative, maybe not, but I think there is great variability in when an artist reaches his or her peak. Some artists may have a series of observations that become very important in their twenties, and don’t change very much over time, and in others there is a constant evolution. I think one thing that is clear is that it takes a while to master whatever artform that someone is working at, nobody picks up a pen and produces a perfect sketch of a face, it takes many, many iterations and practise over many times. I think this is what happens when someone is an art student, they are constantly working on these techniques, making their own observations and getting observations on their work made by teachers.
In disease, people who have never painted, made sculptures, or welded art pieces, suddenly become very interested in the process. Their first works are usually not as good as the ones they produce after they've had the chance to work at a specific media. They do things over and over again, and at some point they start to reach a mastery of their art. So I think there is often a period when they don’t produce something very interesting but there is a drive to do so. That drive pushes them to practise more and more and they reach some sort of a peak, until eventually the degenerative process and injury to circuits causes a loss of their abilities.
So we have this very beautiful but sad story of sometimes art heralding the onset of the degenerative disease process. Soon after the art has appeared the degenerative process gets worse, and eventually the ability to produce art is lost altogether.
AB: Do you think that this drive to produce art arises from disinhibition of certain brain networks, especially in patients who, earlier in their history, were never motivated to produce art? In other words is this artistic ability unveiled and perpetuated by the neurodegenerative process itself?
BM: I do. I think the fact that they never produced art before means that the circuits involved in this process had not been activated. Something about the degeneration, for reasons that we don’t completely understand, leads to an interest, an activation, an actual physical drive to carry out the artistic activities. The theme has been that degeneration on the left side of the brain (language based regions) releases functions on the right side, which are more visual.
AB: Have there been any fMRI studies done in these patients with relation to newly developed artistic abilities?
BM: There is quite a bit of fMRI data that we have collected on our artists. We are in the process of analysing that, but we don’t yet have a coherent story. We wrote about it. William Seeley did these analyses on a woman (Anne Adams) who became a visual artist in the setting of a non-fluent aphasia, and she showed on a blood flow scan increased activity in the right posterior brain region, and actually during that time an MRI was done and she had increased volume in that same area.
There are a number of theories, one being she was always like that (that is the bigger volume). But she was never much of an artist until the progressive aphasia emerged. We think there might have been slow remodeling in the early stages of the disease, with decreased activity in the left frontal insular regions allowing increased activity on the right posterior parietal area and actually some increase in volume.
AB: Does art created by people with brain disease or damage provide insight into brain anatomy and function? Could you provide a few examples?
BM: Surely Anne Adams was a paradigm shift for me to describe the phenomenon of art and dementia, but I had never really thought too much about the mechanism. But because she had undergone an MRI just before the onset of dementia, this really allowed us to look into the circuitry and mechanism. This also allowed me to broaden my thoughts about the topic, so seeing patients who had gardens with beautiful details, flowers, patterns. This is another form of visual creativity that I have become aware of.
AB: As a practising neurologist, how has your encounter with art influenced or changed your own conception about how the brain functions? Do you have specific examples? Did you have to overcome difficulties to promote this field?
BM: I think it has really humanised my approach to patients. It makes me realize that even though dementia is a relentless process, there are many pockets of preservation, and sometimes enhanced function. It is critically important that we recognize this in our patients. It is helpful in diagnosis. What is preserved is telling us something about where in the brain the bad molecules are not accumulating. But it also allows us to think about the patients, about things that are important to them, and help design programs for them and have activities that are meaningful. If you have lost your visual spatial function profoundly, then probably working in art is not going to be satisfying. But if instead there are other areas that are preserved around music or singing or something else, these things have to be kept in mind while thinking about the future for the patient and their families.
I think this should be a routine part of our diagnostic process, that is not only what are the weaknesses, but what are the strengths, and has anything new emerged that is actually a new strength. We do this regularly now at UCSF (it has opened up a whole new side to the evaluation). This also makes me appreciate the unbelievable effort that every patient that we see is putting into their life. When blocked in certain domains, they activate others and use others. So I think about patients in a very different way since the story of art emerged. I think, to a fault, neurologists have often thought about deficits a lot, without really seeing the whole human being, and I think this has really forced me in a very good way to think about the entire human within the ecosystem that they live and interact with others, and some of the things they perceive that might be very important.
AB: I have had the good fortune of discussing both art and neurodegeneration on various occasions with Bruce. Not only do Bruce and I share similar scientific curiosities with regards to art and dementia, I have also found him to be an excellent mentor. He has taught me to follow my heart in the quest to figuring out the brain, and for this I shall be forever grateful!
“I think the next philosophers, the philosophers of the 21st century, are going to be neuroscientists.” - Bruce Miller
The OHBM is dedicated to understanding the anatomical and functional organization of the human brain using neuroimaging. But how to best use brain-activity measurements, including human neuroimaging, to understand computational mechanisms remains an open problem. “Mapping the brain does not by itself reveal the brain’s computational mechanisms” says Niko Kriegeskorte, past chair of the OHBM Communications Committee. “Therefore one of the strategic priorities in the OHBM Communications Committee has been to explore the interaction between computational neuroscience & human neuroimaging.”
Here, we had the chance to discuss the current state and future of computational neuroscience with Mark Humphries, senior research fellow at the University of Manchester, Chair of Computational Neuroscience at the University of Nottingham, and talented blogger. We found out about research environments in different countries, mindful language use in neuroscience, Mark’s outlook on the future of network neuroscience, and his top three tips for those starting out in computational neuroscience.
Nils Muhlert (NM): Can you tell us a bit about your career path - were you first interested in computing, or in neuroscience? Also, your work has seen you move between the UK and France - have you found different approaches to research in these countries?
Mark Humphries (MH): I’m of the generation that grew up programming their home computers - their C64s, Spectrums, and BBC Micros - so computing was always there. As a kid I also loved chemistry. Originally I wanted to do Chemical Engineering at university, but it turned out that A-Level Chemistry was both hard and boring. So when I came across the mysterious “Cognitive Science” degree, promising computing, AI, and the brain, I signed up like a shot. In effect, I’m one of the few who was trained in computational neuroscience from my first year at undergraduate level.
That degree was followed by a PhD and postdoctoral work at Sheffield, with the quietly wonderful Kevin Gurney. Not quite the straight run it sounds: disillusioned and exhausted by the end of the PhD, I went off to freelance web design and software engineering. That lasted a year before I was tempted back by the offer of a post-doc.
My long stint at Sheffield was followed by three years in Paris at ENS. Both teams of computational neuroscientists, with radically different approaches. Sheffield were neuroscience-first, circuit modellers: build a model of a brain region, study its dynamics, and infer its function. Paris were theoreticians first: propose and study general principles for how computations could be done by the brain (memory, inference etc), then worry about the details of specific circuits later, if at all.
In my experience, the French research system, dominated by the CNRS and INSERM, is essentially just part of their civil service system. So you can have a job for life, but getting financial support to do your research can be an absolute pain. Theorists in all fields can thrive, of course. (ENS has an extraordinary maths department: the Bourbaki group were based there, and they’ve had five Fields medalists). The UK research system more clearly supports fundamental science.
NM: In a recent blog post on connectomes, you highlight some of the many factors influencing the spiking of a single neuron. In human neuroimaging, we typically summarise activity at the scale of cubic millimetres, with each voxel containing tens or hundreds of thousands of neurons in different cortical layers. How much cross-talk do you see between cellular systems neuroscience and human neuroimaging, and how much do you think understanding at one level currently constrains understanding in the other?
MH: The neuroscience of detailed neuron types - their physiology, receptors, transmitters, gene expression, and so on - often has little constraint on systems neuroscience studies of large populations of neurons. Many multi-neuron recordings from cortical regions can only hazard a guess at what layer they are recording in, never mind whether the recorded neurons are Martinotti or ViP interneurons or whatever. I think this lack of identifying neurons has played a large role in driving the take-up of calcium imaging, where we can at least identify some subtypes of neurons (typically 1 or 2), despite the obvious disadvantage of recording something (calcium) that is only partially related to the thing we’re interested in (the spiking of neurons). What’s particularly missing is the constraints of anatomy - the wiring between individual neurons - on the activity we’ve recorded from those neurons.
But that will come. In a handful of specialised circuits, this information is being combined. For example, in studies of the mouse retina, the type and position of neurons has been used to constrain classifications of large population recordings. And in tiny animals, like Drosophila larvae (maggots to the rest of us) and C Elegans, the details of wiring and neuron types have been combined with large-scale imaging to reveal deep insights into how brains could work.
NM: Marsel Mesulam revealed that students requesting higher field strength MRIs are asked “what would you do if you could record from every neuron in the brain?” This thought experiment is now an ambition for international research projects. How do you feel network neuroscience could sensibly use this massive amount of data?
A question that has occupied much of my thinking, but to which I’m no closer to a good answer. We have passed the milestone of recording every neuron from a simple nervous system. But as I wrote at the time, it was a cool study from which we learnt very little of consequence.
That said, everything that brains do, they do through the collective action of hundreds to millions of neurons. And we lack well-established theories for what that collective action means, or how to interpret changes to it. In the absence of theory, the gotta-catch-them-all philosophy of recording every neuron is seductive: let’s get the data we think we will need one day, and wait for theory to catch up.
Fortunately, ideas are emerging about how we can sensibly use this data. There’s some great recent work on how we can tell whether there’s anything special about the joint activity of many neurons: whether it is just the expected result of lots of individual neurons tuned to different properties of the world; or if the joint activity really conveys more information than the individual neurons summed together. And we’re starting to get a handle on how to understand the dimensionality of that joint activity: how much redundancy there is between neurons, how that redundancy differs between brain regions (and between different brains), and what that means.
NM: In another of your blog posts, you criticize media misinterpretations of dopamine as representing the ‘reward system’ of the brain. How does your own work feed into this - and at what point did you feel a general education piece was warranted?
MH: The tipping point was seeing “Dopamine dressing” in The Guardian‘s Style section. As though dopamine neurons give a damn about what you wear. Endless publications call dopamine the “reward system”, when it is not. And it’s particularly embarrassing when such language routinely appears in august publications like Nature. So I thought that it’d be useful for everyone to have a simple, accessible, concise explanation that dopamine neurons signal an error, not reward. And then we can all just point our undergraduates, friends, family, and editorial staff at esteemed publications to that post, and save ourselves the trauma.
Dopamine has been around in my research since the first days of my PhD. For years my work was primarily on the basal ganglia, and the striatum - the massive input nucleus of the basal ganglia - is where the dopamine neurons send their dopamine. So we include the effects of dopamine in all our models. In Paris I spent a couple of years analysing dopamine neuron firing in a project that never saw the light of day. More recently, I helped Kevin Gurney achieve his mammoth computational account of how dopamine teaches the basal ganglia to select actions. Dopamine has haunted me for my entire career...
David Mehler (DM): Richard Feynman used to stress the difference between “Knowing the name of something and knowing something”. In a similar spirit, you have critically assessed whether we put too much faith in named brain structures, giving examples why these should not be taken at face value. What advice do you have for students and ECRs, whose experience of Neuroscience may consist wholly of learned brain regions with set functions?
MH: Read more than just about your brain region. And internalise the idea of degeneracy: brains have many solutions to the same problem.
If we work on only one brain region, it is easy to fall into the trap of thinking that one brain region does everything. Just being aware of the thinking about brain regions other than your own will help not take anything at face value. In my own fields, it is easy for basal ganglia researchers to fall into the trap of claiming that it is responsible for “action selection”. But this patently can’t be true: there are multiple systems that select actions in the brain, from spinal reflexes, up through the brainstem, midbrain, and other sub-cortical structures - the amygdala can select fear responses just fine on its own.
DM: A recent study from your lab, in collaboration with Angela Bruno & Bill Frost from the Chicago Medical School, provides fascinating insight into how neural populations orchestrate their activity when coordinating movement: while their combined output converges to a similar pattern (an attractor), activity of individual neurons is not stable over time. What does this finding imply in your view for our understanding of functional connectivity (e.g. between neurons or neural populations)?
It means that functional connectivity is an epiphenomenon. The correlations between individual neurons are imposed by the dynamics of the whole circuit in which they reside. Those dynamics obey certain properties that emerge from the wiring of the whole circuit and the excitability of the individual neurons.
But it is very useful to study functional connectivity of neurons: mapping the correlations between neurons is so much easier than trying to infer the underlying attractor, or other form of dynamical system. And changes to those correlations imply a change to the underlying attractor. Indeed, we use this approach all the time. We just need to be mindful that those correlations are a read-out, an observable property, of the circuit’s dynamics.
Functional connectivity at the level of whole brain regions, of MEG/EEG and fMRI, is a different kettle of fish, of course. On this scale, correlated activity is telling us something about the distribution of how things are represented across the brain in very large neural populations, with tens of thousands to millions of neurons in a single time-series. Instability of correlations over time for these time-series would suggest entire neural populations that wink on or off as needed. And dynamical systems analysis has long been applied to EEG data, but usually as a way of looking for changes in gross neural activity - as may precede an epileptic seizure, for example - than as a view of how the brain computes.
Seeing a spiral attractor in neural activity. Activity was recorded from 105 neurons in a sea-slug's motor network during three separate bouts of galloping. There are three lines plotted here. Each line is the low-dimensional projection of those neurons' joint activity during a 90 second bout of galloping, from its onset (grey circle). Each line traces a circular movement whose amplitude decays over time: a spiral. The three lines together trace the same region of this low-dimensional space, indicating that the neurons' joint activity is attracted to the same pattern: the spiral is an attractor.
DM: Your work increasingly focuses on dynamic changes in neural networks. What insight do you think this will bring to the field over the next 5-10 years?
MH: We’re going after the idea that the brain encodes information at the level of the joint activity of populations of neurons. In this view, each neuron is a read-out of the joint activity of all the neurons that project to it. That neuron, in turn, is just one small component of the populations projecting to other neurons. So only by looking at the dynamics of the neural network as a whole can we understand what neurons are seeing, and hence what the brain is encoding. A change to those joint dynamics are then the change in what is being encoded: be it a sound, a memory, or a movement. In short: the response of single neurons may be irrelevant to what the brain is doing.
DM: … and finally, computational neuroscience is gaining increasing popularity. But starting out may seem daunting. What are your top three tips to get into the field?
MH: First, learn to code, properly. To some, this may seem obvious. In my experience most people who’ve come to me with a genuine interest in getting into computational neuroscience have never coded, certainly not seriously. But coding is the day-in, day-out life of the computational neuroscientist, so you won’t get far without deep skills in coding. And by “properly” I don’t mean “you have to learn a proper programming language”, whatever that means. No: properly learning to code means learning the logic of how code is built, independently of the language used: of variable types, indexing, functions, control loops. And learn to comment your code. You know who will love you for commenting your code? You, in a year’s time.
Second, ask yourself: What type of computational neuroscience do I want to do? The choices are endless. We can work on scales across the actions of receptors at single synapses; plasticity at single synapses; the intra-cellular signals triggered by receptor activation; the dynamics of a single neuron in all its glory, dendrites and all; the collective dynamics of networks of neurons; of specific brain circuits; right up to the entire brain. And on to read-outs of mass activity, to EEG, MEG, and fMRI, and the functional connections between regions. We can work bottom-up, top-down, or middle-out. We can aim to ask what a specific brain regions does, work out what causes a disorder, or reach for general principles for how neurons compute. We can use algorithms, like machine-learning; simulations of dynamics using differential equations; or pencil and paper to solve equations. What is it you want?
Finally, take a Master’s course in computational neuroscience. Both so you can find out if this path is for you; and so that you can be taught the neuroscience by neuroscientists and the computation by computational neuroscientists. Get either wrong, and no one will take you seriously.
By Elizabeth DuPre and Kirstie Whitaker
The open neuroimaging community is great and growing every day. This month saw the first of a series of Open Science Demo Calls. Brought to you by the OHBM Open Science Special Interest Group, these live streamed calls are a chance to hear from the developers of open neuroimaging tools. We'll use these calls to build connections between all members of the OHBM Open Science community and to tell the stories of the people making outstanding and reproducible neuroscience happen.
For our first call, we spoke to Alejandro de la Vega, Cameron Craddock, and Guiomar Niso about three ongoing initiatives they’re spearheading to improve openness in neuroimaging research.
Alejandro spoke about NeuroScout, a new, cloud-based platform allowing for the flexible re-analysis of neuroimaging datasets with naturalistic stimuli, such as the Study Forrest dataset. To do this, Alejandro is actively working to develop tools such as pliers and pybids. If you’re interested in this line of research, make sure to check out and contribute to these tools!
Cameron discussed this year’s Brainhack Global. Building off the successes of Brainhack Global 2017, Cameron is organizing a globally based hackathon for this spring, where neuroimaging researchers around the world can come together online to learn about, develop, and improve open neuroimaging tools. He encourages anyone interested in attending the event to join the Brainhack Slack team.
Technical difficulties prevented us from seeing Guiomar in our call, so we recorded a supplementary video to hear more about her work with MEG-BIDS. This is a very big extension of the BIDS specification to cover MEG data. As Guiomar informed us, MEG does not have a standardized acquisition file format (like MRI dicoms), so the creation of an MEG-BIDS standard will make a huge difference to the community! Feedback is welcomed on the current draft of the specification, which is planned for release on February 14th.
Our next call will be on Thursday February 22nd at 7pm GMT (check your local time zone) and will feature Anisha Keshavan on Braindr, Yaroslav Halchenko on DataLad and Athina Tzovara discussing how research treats underrepresented minorities.
If you’d like to nominate yourself or someone else to be featured on these monthly calls, please add their information at this github issue, or email the host of the calls Kirstie Whitaker at firstname.lastname@example.org. You can join the OSSIG google group to receive reminders each month.
Professor Aina Puce is the Eleanor Cox Riggs Professor in the department of Psychological and Brain Sciences at Indiana University, Bloomington, and a senior editor at Neuroimage. She has followed a career path that is now becoming more common in human brain mapping, starting firmly rooted in the methods end but, over time, gradually shifting focus towards understanding complex patterns of behaviour. To do this, she has made use of a number of imaging techniques, exploring ways to extract converging lines of evidence.
Here, we find out how her interests changed throughout her research, the promises and pitfalls of multi-modal imaging, and why you should not be discouraged by rejections but instead focus on and be motivated by the paper acceptances and other highlights in your career.
Nils Muhlert (NM): You initially graduated with degrees in Physics/ Biophysics. Now, one of your lab’s key interests is specific applications - such as understanding social cognition - though clearly facilitated through your expertise in imaging methods. Can you tell us about how your research focus has changed throughout your career?
Aina Puce (AP): My undergraduate degree was in Biophysics and my Masters degree was in Physics. For my Masters I was already recording EEG/ERPs in the operating room under anaesthesia – generating a frequency response of the visual system using sinusoidal visual stimulation through closed eyelids. During my PhD, I recorded intracranial EEG/ERPs from the hippocampus and temporal lobe for the purposes of identifying the epileptogenic temporal lobe in presurgical patient assessments.
My interest has always been tied to the relationship between brain and behavior. Over the years it has evolved from consciousness under anesthesia, to hippocampal integrity, to recognition memory of objects, to face perception, to recognition of face, hand and body actions, to multisensory perception, and now to the implicit recognition of emotions and other non-verbal signals. Seems like a lot of topics perhaps, but the evolving theme is how we make sense of our world. I owe a lot to my colleagues from the humanities: over the years they have patiently taught me so much about psychology.
NM: Much of your work involves imaging across modalities. Alongside the higher temporal and spatial precision, multi-modal imaging often involves the challenge of combining very large datasets. How have you got round these issues?
AP: Important question. When you study brain function using only one imaging method you will look at the world with a set of (rose-colored) glasses that give you only part of the story. We tend to forget that. Using multiple methods (either across or within subjects) keeps you honest, as you might get different answers to a scientific question. Then the onus is on you to get to the bottom of those differences, which means taking more time to study a problem. This can be frustrating, because at times you feel you are not getting anywhere relative to others in the field. At the same time, I would rather generate work that is reproducible and replicable by others! The field needs a solid foundation, and this can only be achieved by paying attention to data quality and also fully understanding the methods we work with.
With respect to large multimodal datasets, the biggest challenge right now as I see is data quality control. Data will likely be analyzed by individuals who may not have expertise in data acquisition and artifact recognition/rejection. When multiple assessment modalities are involved, this problem becomes compounded.
Another challenge that I see relates to cloud computing and subject privacy. Increasingly, subjects in these big datasets will be patients. As more investigators around the world interact with these datasets there is an increased potential for hacking and accessing sensitive information. Having easy to use, but secure, user interfaces and procedures for interacting with big datasets is key.
Another critical component is user training on computer hygiene. I am continually horrified by what I see those who are not computer-savvy doing with data-archiving and sharing. We cannot blame these people as they have not been formally trained in this area, but these are the potential weak links in the chain. That said, user-training needs to be made meaningful and interesting and something that users view as important – and that is also a big challenge in my opinion.
NM: Where do you see multi-modal imaging going over the next 5 years?
AP: With respect to methods and scientific practice: these have been re-examined and will continue to do so. With respect to neuroscience in general: I think that meso-scale neural interactions will be a major focus, as this work is critical to building bridges between systems neuroscience and molecular/cellular neuroscience.
Finally, for social neuroscience, measuring/monitoring brain and bodily function will also become more important as science moves more and more from a lab-based focus to a real-life one. Smart clothing used with dry electrode portable EEG systems and smartphone applications to gather data will become more common. Exciting developments in MEG sensor technology will continue, with attempts to develop higher temperature MEG devices and also flexible sensor helmets to better fit any head shape or size. This is a really exciting time to be involved in neuroscience!
NM: In your work on social attention you have proposed a ‘socially aware’ brain mode of social information processing. Can you tell us a bit more about this. How, if at all, does this brain mode map onto specific resting-state networks?
AP: I have recently been interested in how we use social information that we access implicitly to make social judgments or decisions about the behavior of others. Most lab-based studies in social neuroscience use tasks where subjects make explicit social decisions about others. Yet, this is so unlike what we do in real-life. In our lab we use both implicit task (involving a ‘default’ mode, where there is an internal focus on achieving goals) and explicit tasks (requiring a ‘socially aware’ mode, where we make explicit social judgements), using the same stimuli in the same subjects. We found very different neurophysiology across tasks – explaining in large part the existing variability seen in the literature.
Relationship to resting-state networks? Excellent question! We have been looking at the EEG dynamics during these implicit and explicit tasks, but have not yet looked at resting state EEG in these same subjects. So this is something that I would like to look at in future work.
NM: What advice would you give an early career researcher to help them stand out in the hunt for competitive fellowships, grants and faculty positions?
AP: I usually tell everyone to find what their passion is. What topic of study really motivates you scientifically? Doing science is a perpetual set of ups and downs – often more down than up. If you follow that passion, you are more likely to be successful, because it will help you get through the bad times.
As for specific advice for early career researchers. First and foremost, find a mentor – a senior scientist who you trust, have a personal rapport with, and who can help you work on your desired career goals. They should be a good sounding board, but also be able to network you with other scientists and point out career opportunities you may not know about. OHBM has an excellent mentor-mentee matching service. I have been recently assigned to mentor two young scientists, and I am looking forward to interacting with them on-line and face-to-face at the OHBM meeting itself!
Second, network network network! Don’t be afraid to speak with senior scientists at scientific meetings – not just at your poster, but do it at the various social events. Getting to know someone can allow you to visit their lab (perhaps even on a short stay to analyze some data), and who knows what other opportunities that might lead to? Applying to competitive Summer schools can also give you this opportunity.
Third, seek feedback from peers and colleagues on your fellowship and grant applications. People do not do this enough. That said, it requires being organized – you need to allow time for people to read and give you feedback, so that you can make the edits before the submission deadline. Same thing applies for job talks or conference talks – in our lab no-one does a talk anywhere without doing a dry run first! This rule also applies to me, and I value the detailed and caring feedback I get from my trainees.
Fourth, you can stand out by being yourself – scientifically and personally. Scientists are by nature prone to eccentricities. I like to celebrate those. Your (hopefully positive) eccentricities make you who you are, and importantly make you distinctive and memorable to others. (I'll never forget a job candidate who told us that he had a pet tarantula. He got the job!)
NM: Next month, you’ll be a keynote speaker at the Brain Twitter Conference. Can you give us some insight into what you’ll be presenting - and what you think can be achieved through this online mini-conference?
AP: I will keynote tweet on the different modes of social information processing that I mentioned before.
What can be achieved with an online Twitter conference? A couple of things quickly come to mind. First, the conference builds a greater sense of community, allowing new connections between scientists around the world to be made through interactions generated in response to speakers’ tweets. (It is interesting to finally meet people at scientific meetings that you have been tweeting with.) Second, communicating one's ideas with a series of 10 Tweets makes one distill the absolute essence of the ideas to be presented. It allows the presenter, at least, to work out what is really important in the practice of their science.
NM: When did you become involved in Neuroimage - and how have you seen it develop over the years?
AP: I became a member of the Editorial Board in 2005, a Handling Editor in 2009, a Section Editor in 2011 and finally a Senior Editor in 2013. It has been wonderful to watch our field grow exponentially over the years and to work with so many dedicated and committed people in our NeuroImage family. Back in the early 1990s we had no outlet where (f)MRI-related work was welcomed, whereas work related to MEG and EEG was being published in well established neurophysiology journals. Today we have NeuroImage as well as Human Brain Mapping (which also began very early to meet the need to publish MRI-related work). It is terrific to see neuroimaging work so mainstream and regularly appearing in high-profile neuroscience journals. Indeed it is hard to keep up with it all right now!
NM: ...and finally, you’re currently serving on the program committee for OHBM. What does this role involve - and how can others contribute?
AP: OHBM is my tribe. As a post-doc I presented a poster at the very first OHBM meeting in Paris organized by Bernard Mazoyer in 1995. I have only missed a couple of OHBMs since then, due to issues related to visas... I have presented in Educational Courses, Symposia and given a Keynote, as well as chairing scientific sessions over the years. I was a member of Council from 1999-2002, where I was the Meetings-Liaison. Back then we did not have the wonderful Secretariat we have now, so the meeting organization was a bit different. Currently, together with Cyril Pernet I am Co-chairing a COBIDAS for MEEG committee for OHBM. I am also a member of the OHBM Scientific Program Committee – and right now this is busy time for the committee. I want to give a huge shout out to Michael Chee and his very capable team in Singapore. World-events forced the change of the meeting city at the last minute, and Mike and his team are making sure that OHBM 2018 will be just as successful as all of our other meetings. I am really looking forward to it!