Professor Francisco Castellanos The article  that Edmund Sonuga-Barke really convened was   an interesting experience, because he really  invited a number of us to participate with him   in taking stock of where the field is,  looking back to where we’ve come from,   and anticipating where we might be going, which  is always the whole reason for doing this. But   it was really an interesting process to do  this in an iterative and collaborative way. So, we think that – again, it’s always  a snapshot in time of when you make your   best guesses as to where things are going. But,  looking back, we can be a bit more insightful   about the optimism that we had at one time,  which may turn out to have been a bit naive,   but, nevertheless, has led us here, so it’s  a looking back and looking forward [pause]. So, I participated in the section that  really focused on our understanding of   the brain in relationship to ADHD. And Jeffrey  Newcorn and I collaborated on that section,   from different perspectives. Mine, more  focused on magnetic resonance imaging,   as the primary approach, and Jeff looking at the  pharmacology and the various methods that are   particularly suited for that, including  positron emission tomography [pause]. So, it’s a humbling process to really take  stock, after a few decades of working fairly   intently to see what really holds up. And  after approximately three decades of work,   we do have some findings that we feel are fairly  firm, in terms of conclusions we can make about   the brain and ADHD, but surprisingly  few. They tend to be relatively global,   in the sense that one of the firmest results is  that, on average, groups that are defined by the   characteristic of having ADHD, and a diagnosis  of ADHD, tend to show smaller brain volumes,   smaller surface area, in general, a decrease  in size, relative to comparison samples. That has to be put in the context of a greater  amount of diversity and variability, so that   it’s not possible to use any of those findings or  measures as an indication that someone has ADHD.   It’s more that, as a group, the averages tend to  be reduced in the direction of less brain matter,   or less neuronal mass. And so, that’s been  something that’s held up, and that’s something   that we found, and the group that I worked with,  found from back in the mid-90s, and it’s been   confirmed in a number of different ways, and so,  I think that that’s about as firm as results get. The next question has always been, what else  is different? Are there specific findings that   can be really related to ADHD, per se? And that’s  where it becomes much more challenging, in part,   because we’re finding that psychiatric disorders  tend to be much more alike than different. And so,   there’s a great deal of overlap across disorders  that present primarily in childhood, such as ADHD,   or those that present in adolescence  or in later life, such as depression,   or even schizophrenia, bipolar disorder. And so,   it’s turned out that there may be common patterns  relating to general psychopathology, which, again,   are easier to discern and less clearly  related to single, individual disorders. Having said that, I’ll come back to, you know, to  the challenges that that really has made – that   we’ve had to encounter about that, when we really  touch on the questions of outstanding questions,   or issues of outstanding questions. So, the  implication of prefrontal cortex roughly holds   up. Striatal circuits seem to be implicated,  the cerebellum, and parts of the cerebellum,   are among the most robust findings  when we look at the brain, sort of,   region-by-region. But the brain doesn’t work in a  region-by-region way, and so, we’re increasingly   looking at relationships between brain networks,  or between brain regions creating networks. And that work is proceeding, and there are results  that look intriguing and may well hold up over   time, but the way that is published – the results  that are published in literature at this point,   suggest that those are real and almost evanescent.  The effect sizes are incredibly small, and it so,   it suggests that we’re not doing  this in an optimal way. And so,   it’s a mixed result of saying, “What do we  really know?” We have some results that hold up,   but they’re quite minor, even in samples  of more than 1,000 individuals in a group,   and so, it makes it much more difficult  to think that that’s going to translate   into clinically actionable  insights anytime soon [pause]. So, the field, like any field that has a new  instrument, and neuroimaging as a whole is a,   kind of, new instrument, it’s  been developing over decades,   but especially the advent of  magnetic resonance imaging,   or MRI, really has revolutionised our  ability to quantify aspects of brain   structure and function. And that’s been really  part of our toolkit for about three decades. And so, initially, we start off by doing, you  know, simple minded pursuits and measuring volumes   and the surface areas. And even that takes quite  a long time and quite a lot of effort to do well,   but we’ve, sort of, got that worked out, more  or less, at a very gross level of structure. So,   typically, 34 parcels per hemisphere in  the brain, for example, for the cortex,   knowing that those are highly heterogenous  regions, for example, but that’s the standard   in the field still, in terms of quantifying  volume, area, other kinds of parameters. But, in parallel, and in part due to some of  the efforts that my group has been involved in,   we’ve become much more aware of the  importance of thinking about the brain   in terms of its circuity. And that can be  revealed in a functional imaging manner,   as well as in structural imaging, looking at  connectivity of white matter. And so, “structural   connectivity” and “functional connectivity,”  as these two approaches are termed, has been   where a lot of the action is in psychiatric  disorders in general, and also in ADHD. And so, the notion is that there are aspects  of how well connected brain regions are,   and whether their connections are  increasing or decreasing appropriately,   that are relevant to thinking about the  development and the persistence of ADHD   symptoms and impairments. It’s important  to highlight that all brain connections are   not equal. Some are quite strong and remain  strong. So, the visual system, for example,   is highly connected to other parts of the visual  system in other higher cortical areas, in a sense,   and that appears to remain the case from very  early life, certainly through development. The relationships and the interactions between  prefrontal and other regions and, sort of,   more abstract processing centres, tends to be  something that strengthens over time. And so,   differences in the way in which  that occurs, or relative, you know,   strength versus weakness in some  individuals, may well underly some   of the difficulties that we observe when  we look at behaviours and symptoms [pause]. So, we’ve been doing the work of trying  to understand the brain with tools,   such as in vivo brain imaging, for about  three decades, and it’s been a slow,   foundational process. It’s really  about beginning to understand the tool,   the statistical methods that we need to apply  in order to be rigorous, and that’s improved   substantially. I’m sure that it will continue  to improve even more, based on – by comparison   to where we are now. But that’s been the most  dramatic change in the field, is that we’ve really   come to acknowledge that we were not controlling  appropriately for false positive results. And so,   we need to take a good look at our literature  and see what really is holding up, in part,   because the effect sizes that seemed to be  real are much smaller than we anticipated. And so, with that perspective, we have a number  of efforts that are quite largescale studies,   ABCD, in the US, the Adolescent Brain  Cognitive Development Study, or IMAGEN,   or, you know, other sort of consortia,  in other regions of the world. Europe,   in particular, has been quite active, but there  are longitudinal studies in China, in Singapore,   that I think will increasingly  be relevant to our understanding. And within those studies, the focus is  increasingly on connectivity in its various ways,   and really trying to understand the functional  and structural underpinnings of how the brain   develops. The challenge is that we have  a many-to-many problem. There are – we’ve   talked about ADHD in this conversation, and  we tend to do this in general as an entity,   but we increasingly realise that there  is a number of different ways in which   these symptoms can manifest. So, there’s a  heterogeneity at the clinical level which   is substantial, and there’s tremendous  heterogeneity at the neuronal level. One of the characteristics of biological  systems, highly evolved ones such as the   ones that we benefit from, our bodies, is  a great deal of redundancy. I’m reading a   book right now which talks about the differences  between human systems. In our current society,   for example, no-one wants to pay for a second  kidney. We have everything, sort of, you know,   cut to the bone in terms of paying for  the least possible infrastructure, and so,   when that fails, everything falls apart.  Whereas, the body’s built on the idea that,   you know, you want a setup to be robust, in  case you lose a kidney, you’re still – you’re   fine with the remaining kidney. You lose one  lung, the other one will do just as well. And so, the brain, likewise, has a tremendous  amount of redundancy. We know in the case   of Parkinson’s disease that symptoms don’t  emerge until more than 85% of the dopamine   neurons are no longer functioning. And so,  there’s a great deal of reserve capacity,   and that makes it difficult to identify  where the deficits are coming from,   because it’s likely a mixture of multiple  factors that are slightly less optimal, that   converge and produce some of the problems that  we see at the behavioural and cognitive level. So, we’re a bit, you know, challenged by this, but  to a certain extent, the move and the awareness   of this – the magnitude of this problem, has  really encouraged us to move towards open science   efforts, with large data accumulations, so-called  big data, and prospective designs. And by applying   those, and using and applying the principles of  replicable science, we can increasingly begin to   untangle this complicated relationship between  neuronal factors, developmental variations,   environmental inputs, and the resulting outputs,  which are our behaviours and symptoms, and/or,   you know, different levels of functioning, which  is what leads to the re – to the designation of   someone as having a disorder when  they’re not functioning as well. So, we’re, I think, in a period of continuing  to stra – to scramble a bit, but science is,   fortunately, always able to progress from these  kinds of periods of confusion. We may be – and   I think we’re approaching the need for really  a, kind of, paradigm shift, and some of that is   coming because of the availability of big datasets  and the computational power which increasingly   grows and allows us to have a better handle  on the, kind of, complexity that’s involved.