Neither Miracle Nor Catastrophe

One year one, here’s my #LongCovid update. TL/DR – with many ups and downs, I’ve improved a lot, and am able (with care) to work, travel, and generally enjoy life again. As I write this I’m in Mexico for a CIFAR meeting, sipping a coffee in the Caribbean morning sun. Six months ago, I was wondering – and pretty doubtful – whether anything like this would be possible again. But here I am. If you’re suffering, as I was and as so many still are, there’s every chance that, with time, you will improve too. (Update as of 27/12/22 – I spoke too soon, as always! See end of this post for an update to this update.)

A satisfactory breakfast

The LC symptoms are still with me, like a stroppy child in the back seat on a long car journey. A week or so ago, perhaps to mark the anniversary of my infection, it kicked off again and I had a rough couple of days. What this tells me is that I’m still learning the pacing lesson. Even – especially – when feeling not-too-bad, it’s important to not overdo things, and ‘overdo’ has taken on a new meaning – as living what I used to think of as normal life. I’m also still masking up where I can in busy places, and avoiding stuffy people-crushes whenever I can.

So that’s where I am, and here are a few other things I’ve learned along the way. What follows is my personal (non-medical) opinion, based on my own experience, which in turn has been shaped by the advice of many people with much deeper knowledge.

First, and it bears repeating, rest is so important and is often the hardest thing to do. Hard because many people with Long Covid still need to do mentally and physically demanding work, and because even if you are able to take time off, doing next-to-nothing can be much more difficult than doing things. Don’t ever feel guilty about resting. Doing nothing is extremely difficult, so praise yourself for when you achieve it (though ‘achieve’ still seems the wrong word), and don’t beat yourself up if you don’t. (And beware ‘graded exercise therapy’ – this can be disastrous). As I was reminded the other week, part of the rest strategy is to not push things too far – physically or mentally – even when feeling OK. I’m still steering clear of cardio exercise, but a day of zooming seems just as likely to wipe me out.

Second, social support is equally important. I’ve been very lucky to have the support of many friends & colleagues, both near and far away, who’ve been with me for the journey. They’ve helped guard against catastrophic thinking, been there with food and company when I’ve been laid low, helped take the load off me at work, and much more besides. I am more grateful than I could ever say. 

Third, keep a diary, even if it’s just one line and a mark out of 10 for how you feel. It’s very helpful to be able to track your symptoms over time, to look for patterns and potential triggers, and eventually to see evidence of improvement (this new app might be worth checking out). One of the most frustrating things about LC is its up-and-down nature. I forbade myself from ever thinking “I’ve turned the corner” because every time I did, I’d hit another wall. But over time there’ve been fewer crashes and the crashes are less severe. My diary has helped me recognise this, and sometimes helped me identify some of the triggers. On the other hand, you can go nuts trying to figure out the reasons for suddenly feeling bad. So if/when you do have a reverse, try not to rake over the coals too much. Sometimes, perhaps most of the time, crashes just happen and it doesn’t help to constantly re-examine might-have’s and should-have’s.

Then there are all the other things. I’m still (mainly) off caffeine and I did stop all alcohol for several months. I was following a low histamine diet quite strictly, but I’m more flexible about this now. Perhaps the trickiest thing is knowing what to do about all the various supplements and treatment options that are out there. Here, the lack of well controlled studies means that we’re mostly all freestyling N=1 “clinical trials” (which of course they aren’t). With that in mind, here are some other things that at least coincided with my beginning to feeling better. Broadly, the things I’ve tried have targeted one or more of the following: (i) reducing (neuro)inflammation, (ii) breaking down potential microclots, (iii) supporting mitochondrial function, and (iv) restoring microbiome balance.

Cold showers (or sea swims). A brief immersion in cold water every morning not only helps tamp down inflammation, it wakes you up – particularly useful for caffeine-free times. The trick is to not think about it too much, and to treat the cold sensations as ‘interesting’ rather than ‘cold’.

For microlots, I’ve been taking nattokinase and serrapeptase since the summer. There’s various anecdotal reports that these supplements help. I take 2x of these each morning, ideally 30 mins before eating. Before breakfast I also take turmeric shots and a slug of Athletic Greens, which has all sorts of vitamins and pre/probiotics, though it is pricey. There’s also evidence for the benefits of antihistamines (& low histamine diet). I take daily over-the-counter cetirizine hydrochloride, and famotidine (a prescription type II antihistamine, mainly active in the gut). I’m taking a bunch of other supplements too, but I don’t have strong feelings for whether they’re working or not. Coenzyme Q10, Rhodiola Rosea, and (flushing) Vitamin B3 may be worth mentioning.

More experimentally, I’ve tried IV infusions of various good things, electrical vagal nerve stimulation, flotation, as well as the staples of acupuncture and meditation. All seem to help a little, but again its hard to say for sure. Looking ahead, I’m intrigued by the potential for low-dose naltrexone to help dissolve the remaining wisps of brain fog (in fact I’ve just started on 0.5mg per day). But there’s a worry with all this. It’s far too easy to go down pricey rabbit-holes, and – given the desperation people with LC feel – to get into financial distress. Be especially cautious about expensive treatments, and don’t overextend yourself financially. This will only make things worse.

Zooming out, Long Covid remains a massive public health disaster, with ~17 million affected in Europe alone. It is a disaster that continue to build, and largely out of sight now that societies around the world have decided that the pandemic is over. It isn’t over. That said, there is hope. When I first started looking for information about Long Covid, back in those most difficult first few months, everything seemed to be either a miracle or a catastrophe. But in my experience, and in the journeys of many I’ve met along the way, a middle course seems much more likely.

Many people (not all) do get better, not quickly, and maybe not fully, but they do improve. With more research into the heterogeneity of Long Covid and into the efficacy of different treatments, the odds will improve for everyone. What’s more, the increased attention to post-viral conditions generated by Long Covid may finally help those with other such conditions, like ME/CFS, who have struggled for decades without adequate support.

For me, Long Covid at its worst was like waking every day feeling like I’d been poisoned. Twelve months later, although I now longer feel this way, I am aware that the symptoms could roar back at any time, and I do not expect to ever fully return to how I was before. Accepting this is challenging, but it’s part of the road that travels between miracle and catastrophe.

UPDATE 27/12/22: Well dammit, I spoke too soon – always a danger with LC. After feeling pretty well for much of my time in Mexico, I started to feel ill again on the way home, on Christmas Eve. Initially it felt like a bit of mild food poisoning, and I was hopeful that a good night’s sleep would see me right. But no. Since Christmas Day I’ve been incapacitated again, struggling to get around the house, and with a near full-house of LC symptoms gleefully bounding around my body and mind. It’s hard to know what the trigger was, indeed whether there was any trigger. But I now need to re-evaluate the days and weeks ahead, as I did so many times in 2022. I am optimistic I will feel better again. I just don’t know when. Much love, sympathy, and renewed enpathy to all others out there navigating Christmas through the toxic fog of LC.

Postscript: If you’re still here, and are interested in the mind and brain as well as Long Covid, please consider taking part in our new study of ‘perceptual diversity’: The Perception Census. You’ll help us advance research into how we each experience the world in a unique way, and you’ll also discover more more about your own powers of perception! Give it a shot. It’s fun, I promise.

Mixed feelings about a hard problem: review of The Hidden Spring

For thousands of years people have wondered about the mystery of consciousness. How can anything made of physical stuff – a brain, for instance – be identical to, or give rise to, a subjective experience? Despite a revival in the …

What’s up with the rubber hand illusion?

What’s up with the rubber hand illusion? Here we clarify how our recent studies should be interpreted, and what they mean for experimental studies of embodiment.

Stepping up

What started in Minneapolis has swept across America and now spreads around the globe. This moment is delivering a new opportunity to deal with an old enemy, to begin setting right many old wrongs. One thing that has become abundantly clear is that it is not enough to whisper support from the sidelines, to take solace in the self-assessment that ‘I am not racist’ and then sit back to see what happens. Each one of us must look to what we can do to make things better.

I am a Professor of Neuroscience in a medium-sized University in England. My research focuses on the neuroscience of perception and of consciousness – the science of how we experience the world (and the self), and of how ‘experience’ happens at all. These questions are deeply fascinating and have many practical implications. I am lucky to have a career that revolves around such interesting topics. I am keenly aware that this luck has not been equally available.

In my research group of nine, there are no Black people. There have been no Black people among the 22 doctoral students and postdoctoral researchers I have mentored since arriving at Sussex University more than a decade ago. There are no Black people on the editorial board of the academic journal I oversee – a board consisting of 29 researchers from around the world. I am co-director of an international research program: the CIFAR program on Brain, Mind, and Consciousness has no Black people among its 34 Fellows, Advisors, and Global Scholars. This is not acceptable. Unfortunately, it is also entirely normal for neuroscience as a field. When wandering the halls of the Society for Neuroscience Annual Meeting – which regularly draws about 30,000 people – seeing a Black person is always a rare event. This is wrong.

Beyond its exclusionary effects, this systemic bias has a pernicious influence on how neuroscience is done. In studies of brain disorders in America, minority groups make up a tiny percentage of the cohorts on which studies are conducted (less than 5%, according to 2018 figures from the Lieber Institute for Brain Development). When neuroscience excludes Black researchers, neuroscience neglects Black people.

How did we get here? For those of us in comparatively senior positions, it is tempting to put it all down to a lack of qualified Black applicants, and to the scarcity of Black researchers in neuroscience more generally. There is some truth to this, but simply re-describing the situation does not solve the situation.

What can be done? There is an urgent need to encourage and support Black students and researchers at all stages of their education, training, and professional development, through scholarships, mentorships, networking events, and so on. Here, organisations such as the Society for Black Brain and Behavioural Scientists are doing excellent work. But progress will be too slow if action is left only to those who have a direct stake.

I will make a commitment. Recognising that exclusion starts early, I will make time to mentor and advise Black students who are keen to find a way into cognitive neuroscience. The opportunities we take are defined by the opportunities we see, and having a personal connection into a new world can make a real difference. If anyone wants to take me up on this, all you have to do is email me here – include ‘stepping up’ in the subject line.

Besides this, wherever I have a leadership role I will develop strategies to encourage greater participation from and representation of Black people, extending the active programmes that already exist to promote equity, diversity, and inclusion. I will also sign up for bystander intervention training – a step that many of us can take to make sure we are not left useless on the sidelines when something goes down.

I was a brown-skinned boy growing up in white rural Oxfordshire. I’m not a stranger to racism. I had hoped that things would get better, that the future would naturally tend towards diversity and inclusivity, to the benefit of all. But I now understand that history doesn’t write itself.  It’s time for each of us to do what we can to make things better.

 

 

Registered Reports now available in Neuroscience of Consciousness

I was in Dublin last week, for the biannual meeting of the British Neuroscience Association.  Amid the usual buzz of new research findings – and an outstanding public outreach programme – something different was in the air. There is now an unstoppable momentum behind efforts to increase the credibility of research in psychology and neuroscience (and in other areas of science too) and this momentum was fully on show at the BNA. There was a ‘credibility zone’, nestled among the usual mess of posters and bookstalls, and a keynote lecture from Professor Uta Frith on the three ‘R’s and what they mean for neuroscientists: reproducibility, replicability and reliability of research. The BNA itself has recently received £450K from the Gatsby Foundation to support a new ‘credibility in neuroscience programme’. Science can only progress when we can trust its findings, and while outright fraud is rare, the implicit demands of the ‘publish or perish’ culture can easily lead to unreliable results, as various replication crises have amply revealed. Measures to counter these dangers are therefore more than welcome – they are necessary.

This is why I’m delighted that Neuroscience of Consciousness, part of the Oxford University Press family of journals, is now accepting Registered Report submissions. Registered Reports are a form of research article in which the methods and proposed analyses are written up and reviewed before the research is actually conducted. Typically, and as implemented in our journal, a ‘stage 1’ submission includes a detailed description of the study protocol. If this stage 1 submission is accepted after peer-review, then a stage 2 submission can be submitted which includes the results and discussion. The key innovation of a Registered Report is that acceptance at stage 1 guarantees publication at stage 2, whichever way the results actually turn out – so long as the protocol specified at stage 1 has been properly followed. Also important is that RRs do not exclude exploratory analyses – they only require that such analyses are clearly flagged up. Of course, not all research will be suitable for the registered report format, but we do encourage researchers to use it whenever they can. I’m very pleased that we have a dedicated member of our editorial board, Professor Zoltan Dienes, who will handle Registered Report submissions and who can advise on the process.

Registered Reports are just one among many innovations aimed at improving the credibility of research. Another important development is the emphasis on pre-registration of research designs, so that planned analyses can be unambiguously separated from exploratory analyses. This may be suitable in many cases when a full Registered Report is not. Neuroscience of Consciousness strongly encourages all experimental studies, wherever possible, to be pre-registered. This can be quite easy to do with facilities like Open Science Framework and aspredicted.org. Better science can also be catalysed through publication of methods and resources papers, including datasets. Here again I’m delighted that Neuroscience of Consciousness has launched a new submission category – ‘methods and resources’ articles – to encourage this kind of work.

As many have already emphasized, this emerging ‘open science’ research culture is not about calling people out or being holier-than-thou. Like many others, I’ve faced my own challenges in getting to grips with this rapidly evolving landscape, these challenges will no doubt continue, and it’s been uncomfortable contemplating some work I’ve led or been involved with in the past. Collectively, though, we have a duty to improve our practice and deliver not only more robust results but also more robust methodologies for advancing scientific understanding. My own laboratory embraced an explicit open science policy several months ago, setting out heuristics for best practice across a number of different research methodologies. This policy came primarily from discussions among the researchers, rather than ‘top down’ from me as the overall lab head, and I’m grateful that it did. One thing that’s become clear is that lab heads and research group leaders would do well to reflect on their expectations of research fellows and graduate students. One well designed pre-registered (ideally registered report) publication is worth n interesting-but-underpowered studies (choose your n). It goes without saying that these changed expectations must also filter through to funding bodies and appointment committees. I am confident that they will.

I prefer to think of these new developments in research practice and methodology as an exciting new opportunity, rather than as a scrambled response to a perceived crisis. And I’m greatly looking forward to seeing the first Registered Report appear in Neuroscience of Consciousness. Whichever way the results turn out.

*

(Many thanks to Chris Chambers for his advice and encouragement in setting up a Registered Report pipeline, to Rosie Chambers and Lucy Oates at OUP for making it happen, to Zoltan Dienes for agreeing to handle Registered Report submissions editorially, and to Warrick Roseboom, Peter Lush, Bence Palfi, Reny Baykova, and Maxine Sherman for leading open science discussions in our lab.)

 

Intentional binding without intentional action: A new take on an old idea

You press a light switch and the light comes on. What could be simpler than that. But notice something. As the light comes on, you probably have a feeling that, somehow, you caused that to happen. This experience of ‘being …

Taking back control

I was born in 1972, a year before the UK joined the European Economic Community. As I grew up, in rural South Oxfordshire, the idea of being part of a world beyond England helped keep me going, helped me believe things would get better. Half Indian and half Yorkshire, with a name that even I couldn’t pronounce properly, I looked forward to being part of a world with all the beauty and diversity of Europe, a world in which the threat of war and nationalism was receding not growing, war which had taken my grandfather before I knew him, before he knew me.

In 2016, the day after the referendum, I was giving a talk at a New Scientist event in London. I was up first, and began with some words about the sadness I felt about the result. Sadness about the UK turning away from the world with all its opportunities and challenges, and sadness about the national self-harm caused by the lies, greed, complacency, and desperation for power that had brought us to this point, to 52% vs 48%.

Now, despite myself, I am angry.

Apparently, Theresa May is preparing to bring her appalling deal back to parliament for a third ‘meaningful’ vote, running down the clock until there are no options left on the table, until there is no table. The deal on offer has not changed. To call the votes ‘meaningful’ is therefore the most moronic oxymoron I’ve ever heard. There is nothing meaningful in repeating a vote you lose (and lose by massive margins) until you get the result you want.

Of course, this is precisely the logic by which we are told it is unacceptable to go back to ask the people what they think. The people, we are told, have given their instructions, and we are compelled to carry them out whatever the cost. But while May’s ‘deal’ has not changed, the consequences of leaving the EU are now entirely and obviously different from the lies and false promises that people voted on during the referendum itself, in a campaign that is increasingly being revealed as riven with corruption and driven by dubious foreign and economic interests. (And yes, we need our own Mueller.)

To refuse a People’s Vote on the basis of it being a threat to democracy is hypocrisy of lowest form.

There are many other reasons for sadness and anger. The shapeshifting of our politicians as they jockey for personal advantage amid their self-generated chaos. The airtime given to the far-right headbangers stirring up regressive nationalistic passions to deepen the divisions that are already tearing our country apart. The pandering to the Ulster Unionists and the threat to peace in Northern Ireland. The blatant lies coming from the government as they pull votes, add votes, trot out the same garbage about ‘taking back control’, attempt shameless bribes to get their way, and plough on to the cliff edge regardless. The absence of any effective opposition to what is the most disastrous leadership I or anyone can remember. Cameron and his mates fleeing the scene to chillax in Italy or Portugal or wherever. The disenfranchisement of the young, the back-burnering of all the non-Brexit government business that might actually matter, and all the time and money and hopes and dreams already burnt to ashes on the Brexit trash-fire.

It’s time for all this to stop.

Our society was and is unequal and the dominant neo-liberal complacency needed shaking up. But this is not the way to do it. We are more divided than ever, half of us sold lies and promises of an impossible future, the other half increasingly disconnected from and despairing of the direction we are headed. The EU, while not perfect, cannot be blamed. We brought this on ourselves. And now it’s clear that parliament, once something to be proud of, cannot form a majority for anything – at least not without May’s deadline-day gun-to-the-head and the prospective horror show of her deal rising like a zombie until it finally staggers over the line. This would not be a triumph of diplomacy and democracy. It would be a travesty.

It’s time to go back to the people. Let them take back control.

Time perception without clocks

Salvador Dali, The Persistence of Memory, 1931

Our new paper, led by Warrick Roseboom, is out now (open access) in Nature Communications. It’s about time.

More than two thousand years ago, though who knows how long exactly, Saint Augustine complained “What then is time? If no-one asks me, I know; if I wish to explain to one who asks, I know not.”

The nature of time is endlessly mysterious, in philosophy, in physics, and also in neuroscience. We experience the flow of time, we perceive events as being ordered in time and as having particular durations, yet there are no time sensors in the brain. The eye has rod and cone cells to detect light, the ear has hair cells to detect sound, but there are no dedicated ‘time receptors’ to be found anywhere. How, then, does the brain create the subjective sense of time passing?

Most neuroscientific models of time perception rely on some kind internal timekeeper or pacemaker, a putative ‘clock in the head’ against which the flow of events can be measured. But despite considerable research, clear evidence for these neuronal pacemakers has been rather lacking, especially when it comes to psychologically relevant timescales of a few seconds to minutes.

An alternative view, and one with substantial psychological pedigree, is that time perception is driven by changes in other perceptual modalities. These modalities include vision and hearing, and possibly also internal modalities like interoception (the sense of the body ‘from within’). This is the view we set out to test in this new study, initiated by Warrick Roseboom here at the Sackler Centre, and Dave Bhowmik at Imperial College London, as part of the recently finished EU H2020 project TIMESTORM.

*

Their idea was that one specific aspect of time perception – duration estimation – is based on the rate of accumulation of salient events in other perceptual modalities. More salient changes, longer estimated durations. Fewer salient changes, shorter durations. He set out to test this idea using a neural network model of visual object classification modified to generate estimates of salient changes when exposed to natural videos of varying lengths (Figure 1).

Figure 1. Experiment design. Both human volunteers (a, with eye tracking) and a pretrained object classification neural network (b) view a series of natural videos of different lengths (c), recorded in different environments (d). Activity in the classification networks is analysed for frame-to-frame ‘salient changes’ and records of salient changes are used to train estimates of duration – based on the physical duration of the video. These estimates are then compared with human reports. We also compare networks trained on gaze-constrained video input versus ‘full frame’ video input.

We first collected several hundred videos of five different environments and chopped them into varying lengths from 1 sec to ~1 min. The environments were quiet office scenes, café scenes, busy city scenes, outdoor countryside scenes, and scenes from the campus of Sussex University.  We then showed the videos to some human participants, who rated their apparent durations. We also collected eye tracking data while they viewed the videos. All in all we obtained over 4,000 duration ratings.

The behavioural data showed that people could do the task, and that – as expected – they underestimated long durations and overestimated short durations (Figure 2a). This ‘regression to the mean’ effect is known as Vierodt’s law in the time perception literature and is very well known. Our human volunteers also showed biases according to the video content, rating busy (e.g., city) scenes as lasting longer than non-busy (e.g., office) scenes of the same physical duration. This is just as expected, if duration estimation is based on accumulation of salient perceptual changes.

For the computational part, we used AlexNet, a pretrained deep convolutional neural network (DCNN) which has excellent object classification performance across 1,000 classes of object. We exposed AlexNet to each video, frame by frame. For each frame we examined activity in four separate layers of the network and compared it to the activity elicited by the previous frame. If the difference exceeded an adaptive threshold, we counted a ‘salient event’ and accumulated a unit of subjective time at that level. Finally, we used a simple machine learning tool (a support vector machine) to convert the record of salient events into an estimate of duration in seconds, in order to compare the model with human reports.  There are two important things to note here. The first is that the system was trained on the physical duration of the videos, not on the human estimates (apparent durations). The second is that there is no reliance on any internal clock or pacemaker at all (the frame rate is arbitrary – changing it doesn’t make any difference).

Fig 2. Main results. Human volunteers can do the task and show characteristic biases (a).  When the model is trained on ‘full-frame’ data it can also do the task, but the biases are even more severe (b). There is a much closer match to human data when the model input is constrained by human gaze data (c), but not when the gaze locations are drawn from different trials (d).

There were two key tests of the model.  Was it able to perform the task?  More importantly, did it reveal the same pattern of biases as shown by humans?

Figure 2(b) shows that the model indeed performed the task, classifying longer videos as longer than shorter videos.  It also showed the same pattern of biases, though these were more exaggerated than for the human data (a).  But – critically – when we constrained the video input to the model by where humans were looking, the match to human performance was incredibly close (c). (Importantly, this match went away if we used gaze locations from a different video, d). We also found that the model displayed a similar pattern of biases by content, rating busy scenes as lasting longer than non-busy scenes – just as our human volunteers did. Additional control experiments, described in the paper, rule out that these close matches could be achieved just by changes within the video image itself, or by other trivial dependencies (e.g., on frame rate, or on the support vector regression step).

Altogether, these data show that our clock-free model of time-perception, based on the dynamics of perceptual classification, provides a sufficient basis for capturing subjective duration estimation of visual scenes – scenes that vary in their content as well as in their duration. Our model works on a fully end-to-end basis, going all the way from natural video stimuli to duration estimation in seconds.

*

We think this work is important because it comprehensively illustrates an empirically adequate alternative to ‘pacemaker’ models of time perception.

Pacemaker models are undoubtedly intuitive and influential, but they raise the spectre of what Daniel Dennett has called the ‘fallacy of double transduction’. This is false idea that perceptual systems somehow need to re-instantiate a perceived property inside the head, in order for perception to work. Thus perceived redness might require something red-in-the-head, and perceived music might need a little band-in-the-head, together with a complicated system of intracranial microphones. Naturally no-one would explicitly sign up to this kind of theory, but it sometimes creeps in unannounced to theories that rely too heavily on representations of one kind or another. And it seems that proposing a ‘clock in the head’ for time perception provides a prime example of an implicit double transduction. Our model neatly avoids the fallacy, and as we say in our Conclusion:

“That our system produces human-like time estimates based on only natural video inputs, without any appeal to a pacemaker or clock-like mechanism, represents a substantial advance in building artificial systems with human-like temporal cognition, and presents a fresh opportunity to understand human perception and experience of time.” (p.7).

We’re now extending this line of work by obtaining neuroimaging (fMRI) data during the same task, so that we can compare the computational model activity against brain activity in human observers (with Maxine Sherman). We’ve also recorded a whole array of physiological signatures – such as heart-rate and eye-blink data – to see whether we can find any reliable physiological influences on duration estimation in this task.  We can’t – and the preprint, with Marta Suarez-Pinilla – is here.

*

Major credit for this study to Warrick Roseboom who led the whole thing, with the able assistance of Zaferious Fountas and Kyriacos Nikiforou with the modelling. Major credit also to David Bhowmik who was heavily involved in the conception and early stages of the project, and also to Murray Shanahan who provided very helpful oversight. Thanks also to the EU H2020 TIMESTORM project which supported this project from start to finish. As always, I’d also like to thank the Dr. Mortimer and Theresa Sackler Foundation, and the Canadian Institute for Advanced Research, Azrieli Programme in Brain, Mind, and Consciousness, for their support.

*

Roseboom, W., Fountas, Z., Nikiforou, K., Bhowmik, D., Shanahan, M.P., and Seth, A.K. (2019). Activity in perceptual classification networks as a basis for human subjective time perception. Nature Communications. 10:269.

 

Be careful what you measure: Comparing measures of integrated information

Our new paper on ‘measuring integrated information’ is out now, open access, in the journal Entropy. It’s part of a special issue dedicated to integrated information theory.

In consciousness research, ‘integrated information theory’, or IIT, has come to occupy a highly influential and rather controversial position. Acclaimed by some as the most important development in consciousness science so far, critiqued by others as too mathematically abstruse and empirically untestable, IIT is by turns both fascinating and frustrating. Certainly, a key challenge for IIT is to develop measures of ‘integrated information’ that can be usefully applied to actual data. These measures should capture, in empirically interesting and theoretically profound ways, the extent to which ‘a system generates more information than the sum of its parts’. Such measures are also of interest in many domains beyond consciousness, through for example to physics and engineering, where notions of ‘dynamical complexity’ are of more general importance.

Adam Barrett and I have been working towards this challenge for many years, both through approximations of the measure F (‘phi’, central to the various iterations of IIT) and through alternative measures like ‘causal density’. Alongside new work from other groups, there now exist a range of measures of integrated information – yet so far no systematic comparison of how they perform on non-trivial systems.

This is what we provide in our new paper, led by Adam along with Pedro Mediano from Imperial College London.

*

We describe, using a uniform notation, six different candidate measures of integrated information (among which we count the related measure of ‘causal density’). We set out the intuitions behind each, and compare their properties across a series of criteria. We then explore how they behave on a variety of network models, some very simple, others a little bit more complex.

The most striking finding is that the measures all behave very differently – no two measures show consistent agreement across all our analyses. Here’s an example:

Diverse behavior of measures of integrated information. The six measures (plus two control measures) are shown in terms of their behavior on a simple 2-node network animated by autoregressive dynamics.

At first glance this seems worrying for IIT since, ideally, one would want conceptually similar measures to behave in similar ways when applied to empirical test-cases. Indeed, it is worrying if existing measures are used uncritically. However, by rigorously comparing these measures we are able to identify those which better reflect the underlying intuitions of ‘integrated information’, which we believe will be of some help as these measures continue to be developed and refined.

Integrated information, along with related notions of dynamical complexity and emergence, are likely to be important pillars of our emerging understanding of complex dynamics in all sorts of situations – in consciousness research, in neuroscience more generally, and beyond biology altogether. Our new paper provides a firm foundation for the future development of this critical line of research.

*

One important caveat is necessary. We focus on measures that are, by construction, applicable to the empirical, or spontaneous, statistically stationary distribution of a system’s dynamics. This means we depart, by necessity, from the supposedly more fundamental measures of integrated information that feature in the most recent iterations of IIT. These recent versions of the theory appeal to the so-called ‘maximum entropy’ distribution since they are more interested in characterizing the ‘cause-effect structure’ of a system than in saying things about its dynamics. This means we should be very cautious about taking our results to apply to current versions of IIT. But, in recognizing this, we also return to where we started in this post. A major issue for the more recent (and supposedly more fundamental) versions of IIT is that they are extremely challenging to operationalize and therefore to put to an empirical test. Our work on integrated information departs from ‘fundamental’ IIT precisely because we prioritise empirical applicability. This, we think, is a feature, not a bug.

*

All credit for this study to Pedro Mediano and Adam Barrett, who did all the work. As always, I’d like to thank the Dr. Mortimer and Theresa Sackler Foundation, and the Canadian Institute for Advanced Research, Azrieli Programme in Brain, Mind, and Consciousness, for their support. The paper was published in Entropy on Christmas Day, which may explain why some of you might’ve missed it!  But it did make the cover, which is nice.

*

Mediano, P.A.M., Seth, A.K., and Barrett, A.B. (2019). Measuring integrated information. Comparison of candidate measures in theory and in simulation. Entropy, 21:17