Wednesday, June 24, 2020

how (not) to apply to a phD program in psychology / neuroscience (to work on consciousness)?

it’s that time of the year again. students are starting to think about applying for grad school. over the years i find myself giving some of the same advice to people, and thought i might as well write them down for open sharing. this is mostly from a US perspective, but some of these apply elsewhere too

(why) should one go to grad school?

it shouldn’t be like this, and it is on people like me to fix the system asap … BUT the current reality is grad school pays people very little (~US$30k a year, varies from places to places), and does not offer very good career opportunities after. in the old days, if you’re in a good phD program, chances are you’ll become a professor one day, if you work hard towards that goal. professors aren’t paid super well either, but pay isn’t everything. we have extremely rewarding and meaningful jobs; tenure is a nice guarantee of security & freedom, and an undeniable privilege.

but the old days are gone. you should look up the phD -to-tenure-track (faculty position) ratio yourself. it doesn’t look so good. they say with a phD there are also non-academic career opportunities. that’s correct. but many people also have doubts about whether the years of training are really worth it. again, it is on people like myself to make sure we equip trainees with more transferable skills. but one thing to consider is the sheer opportunity cost of being in grad school for a good half a decade. it can be fun, but it can also be stressful. but above all, you will not be getting the industry training and wages during those years. you get to have your phD, but in sense, you’ll also be missing a head start, if industry is ultimately your destiny. a few years’ head start during the prime of your years is a lot.

the issue is complicated and i don’t draw any firm conclusions here. you have to figure it out yourself. i’m only suggesting you think this through carefully. ask people for advice, but understand too that there may be a bias: if they are already successful in academia there may be a sampling bias re: how good their experiences have been. or even if they aren’t having such a good time, there’s a psychological factor that people tend to avoid being too discouraging and negative. especially if they feel it is their jobs to defend and to promote their discipline, rather than to badmouth it. this kind of sense of loyalty to one’s discipline is understandably common. so, one needs to give it some thoughts re: how to get the hard facts. ask the questions more directly, if you’re unsure. that tends to help.

despite these caveats, i think there is at least one very good reason to apply to grad school: if this is your calling, perhaps you just have to. sometimes in life we choose to do things knowing full well it is hard. maybe the odds are stacked against us. we know we may not be compensated generously, or even fairly. but we just can’t see ourselves doing anything else. or better still - we thought through the other options, and even if they pay better, are more secure, etc, they just aren’t nearly as appealing. that, to me, is one good reason to go ahead and take the plunge. of course, you may or may not agree with this. the above is more like life advice rather than professional advice.

what i said here will probably be misunderstood by some. perhaps it can’t be helped, but let me re-emphasize: i am not trying to normalize the situation, to say that people should be prepared to sacrifice. i totally believe that the system should be kinder to them. just like jazz musicians shouldn’t be expected to starve in order to play great music. i agree, and i am fully aware it is on people like myself to fix this problem for academia. but i also feel that, meanwhile, it’s my responsibility to tell people the truth re: what to expect for now.

how to choose a suitable school & PI?

the first thing is: start thinking about this early. like, ideally, a year or months before your application. ask your academic mentors for advice. multiple of them. cold call people if necessary - many people are actually willing to give free advice to people in your position. they understand this can be a confusing stage to be at. they’ve been there.

the reason why you really need to absolutely think this through, re: whom to work with, is that it will matter a lot for your grad school experience and also your career prospects after. often this factor (PI, i.e. your primary advisor / head of the lab) is much more important than the general prestige of the school / program. unfortunately, some PIs aren’t as on top of things as others. some may be better at placing their students in desirable jobs than others. some may give you more freedom. some may give you so much freedom that there isn’t enough guidance for you - so it sometimes is a matter of style and match too. however, very unfortunately, unprofessional, exploitative PIs do exist, even in the best phD programs. the tenure system means it is not so easy to get rid of them….

... which leads to the interesting consideration of the trade-off between working with someone junior or senior. senior people have a proven track record, and sometimes can offer you better career and research opportunities. junior (i.e. pre-tenured) PIs may lack such experience. but they may be ‘hungrier’. they need to succeed themselves, and thereby they also need you to succeed somewhat. if things don’t go well, it hurts their careers too.

but if the above analysis is right, of course the worst scenario would be a senior PI who isn’t so great to work with. they have tenure, and little to lose. they may have decades of experience dishing out not-so-great experiences to students. and if they made it this far, they gotta be good at (hiding) it. you probably wouldn’t stand a chance against them. (sorry to have to put it in such a scary way!)

so write to the PI. meet and talk with them if they are willing. talk to others about the PI’s reputation. how is their work received by their peers (who will in turn evaluate your own work in the future)? talk to their alumni. what was it like to work with the PI? sometimes people are more willing to tell you things truthfully on the phone rather than through emails. this may sound surprising if not utterly messed up: but people are often worried about retaliation for whistleblowing. of coz, don’t take any one-sided gossips at face value - but when you talk to enough people you should get the overall picture. this can take some work. so do all this well before the interview. i mean, well before you even apply. start now!! just email some people already.

i cannot stress enough how important this is. getting into the wrong lab can be a really, really nasty experience. you can usually get out of the lab (with some considerable hassle), coz the program should protect you. this is one way where it matters who else is on the faculty in the program. so you aren’t choosing just the lab, unless you can be 100% sure you will stay with that lab. (but mind you, great PIs sometimes get job offers and move to different institutions too. sometimes it can happen during your phD. moving along with them can be a slightly tricky business.) but in most cases, if you join a suitable lab in the first instance, you’d be fine. it tends to work out that way.

how to choose a good lab for consciousness?

this is the insanely tricky part. the first thing to realize is, in the US, consciousness remains a ‘taboo’ topic - in a very specific sense: mainstream federal funding for this kind of research is generally lacking. this means you will see a lot of promotion of this type of work in popular books, mass media, internet, etc., maybe even some high profile journals, some theoretical review papers. probably way too much. but actually, there are not really that many well-funded labs doing empirical work on the topic that is respected or taken so seriously by other scientists (see e.g. this, this, or this). and when you do a phD in cog neuro, typically it’s the empirical work that matters. for some solid 'good' labs, even if their research looks somewhat relevant, when you tell them you want to do a phD on consciousness with them, they may think you're being silly

private funding ensures that some of this work continues to be done somehow, but from the top-tier research universities’ administrative perspective, this kind of money is not nearly as good as federal funding (e.g. via NIH / NSF). there’s the prestige factor, but there are also economic reasons. accordingly, there are fewer jobs created in these places for consciousness.

there are two consequences. the first is that you should know what you’re getting into. in terms of intellectual stimulation and challenge, this may be the most fun topic out there. but in terms of career prospects, this may mean there are extra hurdles still. the long and short is, it can be done. many people including myself work on this and manage to not get fired somehow~ if you get in, we’ll share with you our experiences and help you out. but i feel i need to warn you about the reality early on.

another issue is how to choose a lab to start with. lack of mainstream federal funds for this kind of work also means the work becomes relatively unconstrained by the conventional peer review mechanisms. with private funding, sometimes anything goes. together with various socio-historical factors, things can go pretty wild, and at times unhealthily sectarian. so, in addition to all the caveats mentioned in the last section, there is an additional factor for consciousness: if you inadvertently joined a lab where the work is not considered so favorably by other academic colleagues… it may be a situation rather difficult to get out of. depending on the PIs’ style, it may not be in your interest to aggressively challenge their views too much. good PIs should value critics, but, unfortunately some are better than others at this. and if you don’t confront them as much, you may end up doing what they want you to do - then others may see you as being hopelessly indoctrinated into something rather unscientific, due to your own lack of critical thinking, etc. and these people could include your potential future employers too. so this can truly be a career disaster...

so, all i can say is, when it comes to consciousness, handle it with care, seriously. doing a phD, even if it isn’t a sacrifice, is a major life decision. the advice in the last section applies. but for consciousness, don’t just read pop sci / news articles and go with what they say. this is true for most fields, but for our field specifically, the correlation between media prominence and actual scientific quality is often strangely negative. so by all means, do your own research early on. ask as many active researchers in the field as you can. get a diverse set of opinions. we are here to help.

hope you don’t get too discouraged by what i say. if done right, doing a phD in consciousness can be the most wonderful experience too. i say this from my own experience, with all my heart. best of luck!

Hakwan Lau

June 24, 2020


ps - since writing this my lab has decided to do something to help prospective grad applicants to find a suitable lab to work on consciousness in the US. follow me on twitter @hakwanlau. we’ll have something to announce in a few weeks.

Thursday, May 28, 2020

Open letter to NIH on Neuroethics Roadmap (BRAIN initiative) 2019

a while back we sent the below letter. some people ask if they could refer to it. so i thought i could paste it here, so they can link to it. it was an open letter anyway

***

May 14, 2019

Dear colleagues,

Re: Neuroethics needs a balance between theory development and rigorous experimental research on consciousness

We write in response to the call for comments on the Neuroethics Roadmap, which is part of the NIH BRAIN Initiative. We are pleased to see that research on consciousness is receiving recognition. As a group of active researchers in the relevant fields, we hope to point out some potential caveats.

The Roadmap emphasizes the need for theoretical and mathematical models of consciousness. However, current theories are tentative and limited. To make progress, we need experiments designed to identify the neural mechanisms distinguishing conscious from unconscious processes in humans, in whom consciousness can be assessed via subjective reports.

We believe that active research into the neural signatures of consciousness in these relatively clear cases is crucial for building and testing models of awareness in non-human primates and simpler animals, or other more controversial cases. For this reason, for example, it is too early to view the Cambridge Declaration on Consciousness, which states that some non-human animals without a neocortex are conscious, as reflecting a scientific consensus.

Relatedly, the Integrated Information Theory (IIT) of consciousness features prominently in the references cited in the Neuroethics Moonshot section of the Roadmap. While there is support for the association between consciousness and the complexity of active brain networks in humans, some have taken this relationship to generalize lawfully to all physical systems. We do not consider this interpretation to be scientifically established or testable at the moment. We note that alternative theoretical approaches are already referenced indirectly in the Moonshot section (Reference 8); other useful reviews are also available.

Unfortunately, despite their clear relevance to various areas of brain and mental health research, empirical projects focusing on the brain mechanisms distinguishing between conscious and unconscious processes in awake individuals currently do not receive adequate funding. To test theories, we need the relevant data. As many as 58 authors in the field have recently expressed related concerns in a peer-reviewed statement.

Finally, we emphasize that many potential stakeholders could contribute to the Subgroup’s discussions. For example, the Association for the Scientific Study of Consciousness (ASSC) is an open academic society specifically dedicated to scientific research on consciousness. It could have an important dialogue with the Subgroup, and with others interested in the topic.

Thank you for your attention. We have elected to make this letter open, as others may benefit from these clarifications on the current state of research in this area.

Best regards,

(in alphabetical order)

Michele A. Basso, UCLA
Diane M. Beck, University of Illinois
James Bisley, UCLA
Ned Block, NYU
Richard Brown, Laguardia College New York
Denise Cai, Mount Sinai Icahn School of Medicine
David Carmel, Victoria University of Wellington
Axel Cleeremans, Université Libre de Bruxelles
Stanislas Dehaene, College de France
Stephen Fleming, University College London
Chris Frith, University College London
Simon van Gaal, University of Amsterdam
Michael E. Goldberg, Columbia University
Mel Goodale, Western University
Patrick Haggard, University College London
Biyu He, NYU
Sid Kouider, Ecole Normale Superieure, Paris
Robert T. Knight, UC Berkeley
Konrad Kording, UPenn
Hakwan Lau, UCLA
Dominique Lamy, Tel Aviv University
Joseph LeDoux, NYU
Stephen Macknik, SUNY Downstate Medical Center
Susana Martinez-Conde, SUNY Downstate Medical Center
Matthias Michel, Sorbonne University
Lisa Miracchi, University of Pennsylvania
Earl K. Miller, MIT
Lionel Naccache, Sorbonne University, ICM
Adrian M. Owen, Western University
Richard E. Passingham, University of Oxford
Elizabeth Phelps, Harvard University
Megan A. K. Peters, UC Riverside
Dario Ringach, UCLA
Tony Ro, Graduate Center, City University of New York
David Rosenthal, City University of New York
Jérôme Sackur, École des Hautes Études en Sciences Sociales
Yuka Sasaki, Brown University
Claire Sergent, Université de Paris
Anil Seth, University of Sussex
Michael Shadlen, Columbia University
Jacobo Diego Sitt, INSERM-ICM
Catherine Tallon-Baudry, INSERM, PSL Research University
Frank Tong, Vanderbilt University
Peter Ulric Tse, Dartmouth
Takeo Watanabe, Brown University
Thalia Wheatley, Dartmouth

Thursday, May 14, 2020

why i am not a biopsychist

my last blogpost was really meant to be just for fun. but i do mean it for some of the ideas defended, specifically that there are only qualia as relationally understood, wrt to other experiences a person can have. this is to say, there are really no qualia as intrinsic properties. essentially, this means i’ll take something close to what is called a Frege-Schlick view, but will extend and defend it to make it less restrictive. so under some circumstances we can compare experiences across people after all, with the caveat that we can never be completely certain about such comparisons. this will allow us to resist a whole array of challenges to functionalism, e.g. inverted spectrum, zombies, etc. expect a more serious post / draft-y paper in about a month….

***

meanwhile, let’s talk about a different issue, re: how an important debate within the field is taking shape. it was one of those rare moments that something useful philosophically has come out of twitter of all places…

in a tweet i pointed out that panpsychism isn’t really taken seriously by scientists, as some may get that illusion by reading online stuff these days. Victor Lamme challenged me to do a poll. that led to some discussions. turns out, Victor isn’t a panpsychist. surprise, surprise. and in fact, i’m not sure who really is, within the scientific community…. apart from a few *really* far out folks.

but Victor isn’t a functionalist like i am either. by functionalist i don’t restrict myself to specific versions of it; i definitely do NOT think consciousness supervenes on ‘long-arm’ functional properties only. basically, i include for consciousness anything that can be done using software / algorithms. so any computational, representational properties may be relevant. i just don’t think the specific substrate for implementing the software is ultimately that crucial / irreplaceable.

so Victor decided to call himself a biopsychist, i.e. he believes that for a creature to be conscious, the relevant substrate needs to be biological. in fact, he thinks that most if not all living organisms are conscious!

i thought it’s a new name, biopsychism. but very quickly i stood corrected. as Evan Thompson pointed out, the term at least traces back to Ernst Haeckl in 1892.

i really like this way of setting up the debate: biopsychism vs functionalism (broadly construed, as i described above). this really gets at the heart of the issue, a sensible divide within the field, with very legit people on both sides. none of that far out / strawman stuff.

happily Ned Block seems to approve of the term. of coz he’s a biopsychist *of some sort*. he has long argued against functionalism and representationalism. something about the substrate is what really enables conscious experiences. does he think all biological organisms are conscious? probably not. so we should distinguish between some different versions of biopsychism. below i’ll do so, and also highlights some problems i see, even with rather weak versions of biopychism. despite these misgivings, unlike my stance on panpsychism as a scientist, i think biopsychism a position worth taking seriously.

***

the way Evan Thompson defines it, biopsychism states that all and only all living organisms are conscious. that is pretty strong. we can call this strong biopsychism or something. i think it’s quite unlikely to be true. (Evan gave a paper at PSA a couple of years ago, you can email him to ask for a copy; Peter Godfrey-Smith has also written on something related here)

the reason is, i think it’s pretty clear that even simple perceptual experiences involve fairly complicated computational processes that may critically depend on areas of the brain that are matured late in development and evolution, e.g. areas in the prefrontal cortex. a very simple living organism is not gonna have that.

but my take on the empirical matter aside, there is also a pretty damning conceptual problem. so let’s say you’re seeing two different images, a cat and a monkey, in binocular rivalry. when you are consciously seeing a cat, your cat-representing neurons fire. your monkey-representing neurons fire relatively little, as you are not conscious of the image of the monkey. but now, aren’t these monkey-representing neurons biological and alive?

this is related to what is called the 'combination problem', which is something that panpsychists also have to deal with. ultimately, a strong biopsychist will have to say something like, ok, so when you are not consciously seeing the monkey image, the monkey-representing neurons are not signaling consciousness for you. but they are themselves conscious. you just don’t feel their consciousness even though they are in your head.

this leads a rather hilarious way to make Ian Philips happy, i guess: there really is no unconscious perception, as Ian has argued. unconscious perceptual processes are conscious after all, but only becoz everything in the brain is conscious, with or without you~ this is a scenario rather different from the one Ian has argued for, of coz. he’s a reasonable guy. this, on the other hand, seems….. rather weird.

***

so some may hold a weaker version of biopsychism, and say, not all biological organisms are conscious. but if a creature is conscious, it must be biological. the relevant substrate can’t be replaced by something non-biological and yet functionally similar. if you replace it, the subjective experience will be gone, even if the subject behaves somewhat similarly. Ned is likely a biopsychist of this sort.

i am not sure about even this weaker version. because in biology, we look for mechanisms. not magical substrates. let’s say people found that consciousness requires a certain pattern of activity, involving some particular type of neuron, with certain transmitter receptors. then the scientific question to ask is what does it *do*. to the extent we figure out what does it do, why can’t we write down the computational algorithm that would mimic exactly what it does? then we should be able to replace it with something exactly functionally equivalent. if it does the same thing, exactly, and yet consciousness is missing... this just sounds like magic. and how are you ever gonna know?

but Bryce Heubner and Evan convinced me that there may be something to it. the idea is, yeah sure you can try to artificially mimic a biological mechanism. but the mechanism may be so inherently biological, that it involves implementing it in the right bio-habitat, letting it ‘survive’ on its own, do its metabolic work, etc. in that sense, yes, you may try to mimic it, but by the time you succeed…. maybe it isn’t so crazy to say that the artificial replacement is basically just as alive and biological.

i’m still not totally sure, but it’s true that in the old days, we talked about multiple realizability as if it is commonplace. given the same function, we can implement it any way we want. but increasingly, i think people do recognize that multiple realizability is not as common as we thought. often a mechanism can really only be implemented exactly and most efficiently in just one way.

***

so i’m still not sold, but i think it would make an interesting debate. and this may help the field move away from all those distractions we’ve seen in recent years too!

thinking ahead…. i imagine what kind of biopsychic i would become, if i were to eventually come round to it. i suspect the requirement for higher ‘cognitive‘ functions, e.g. those in the prefrontal cortex, is unlikely to give. if this turns out to be empirically wrong, that’s that. but i’m fairly sure for now. if anything, i may come to accept that: if we are to implement those higher functions as the way they work exactly, in conscious creatures like ourselves, yeah, maybe you’d end up having to do something kinda biological. perhaps we can call a person holding such a position a high functioning biopsychist. i’m not one yet, but i got the feeling that my good friend and co-author Richard Brown may be one!