#65 The Mental Sweet Spot: Play, Risk, Surprise & Active Inference with Julian Kiverstein, senior researcher at Amsterdam UMC

Send a love message

 Julian Kiverstein is a senior professor at Amsterdam Medical. He's co-authored various papers about the importance of play as it relates to predictive processing, active learning, intimacy and embodied cognition.

Sometimes we get stuck in attractor states. Play is a form of disruption that may be able to help us get unstuck. Play seems to be far from algorithmic.

This conversation explores how playfulness is crucial for meaning and flourishing, and how building safe spaces for play (such as museums and parks) are crucial for healthy societies.

Active inference and predictive processing are discussed as tools that might help us better model and understand this ‘sweet spot’ towards finding ways to create spaces where we can explore uncertainty and risk without danger.

Drawing from his extensive work in phenomenology, embodied cognition, and predictive processing, Julian offers fresh perspectives on how play connects to mental health and wellbeing. Some key ideas from this episode:

• Play requires safety yet involves taking risks—a paradoxical relationship that enables personal growth
• Adults often lose the curious openness of childhood as we become fixated on seriousness and habitual patterns
• Love shares qualities with play as both involve transcendence beyond the self and openness to fresh experiences
• Active inference and predictive processing provide frameworks for understanding both mental illness and flourishing
• Breaking out of "attractor states" or fixed patterns requires disruption that playful activities can provide
• Creating safe spaces for play becomes essential for development, creativity, and meaning-making

00:00 The Role of Play in Well-being
01:50 Introduction to Love and Philosophy
02:45 Exploring Active Inference and Predictive Processing
05:24 The Importance of Play in Development
09:58 Julian's Journey into Mind Studies
12:11 Understanding Mental Illness through Predictive Processing
21:57 The Concept of Play and Its Cognitive Benefits
30:27 Intrinsic Motivation and the Value of Play
44:12 Play as a Disruptive Force in Mental Health
45:09 Understanding Mental Illness and Uncertainty
46:13 The Role of Play in Mental Health
47:38 Creating Safe Spaces for Emotional Regulation
49:05 Exploration vs. Exploitation in Learning
52:03 The Importance of Play in Adulthood
53:35 Art, Literature, and Emotional Engagement
56:55 The Need for Play in Academia
01:20:50 Balancing Exploration and Familiarity
01:23:37 Final Thoughts on Play and Well-being

Intimate Places: Playgrounds for self-exploration

Julian's papers

JK and Darius Active Inference Institute

Harry Heft Conversation

Karl Friston Conversation

Blog post about Active Inference

4E Cognitio

Support the show

Please rate and review with love.
YouTube, Facebook, Instagram,
question of what is it to bewell?
And for things to go well, youneed to break your own habits,
you need to challenge thosehabits and step outside of your
comfort zone, sometimes when youfeel safe to do so.
That's where you get to grow,to develop, to grow, to develop.
It connects very much with theliterature on active learning

(00:25):
and play and learning, wherethere's this idea that what
children are doing when theyplay is seeking out this sweet
spot where things are not toodifficult or too complex so that
the children don't understandwhat's going on, but nor are
they so simple that the childrenhave nothing to learn.
We kind of lose this curiousopenness over the course of our

(00:47):
development that we start outwith as children.
That gets lost as we becomeadults and we get this idea that
, well, we need to be serious.
There's this constant fallingback on habitual ways of
engaging with the world at theexpense of more openness,
curiosity, exploration.
So I can say cliche thingsabout love, but I don't think

(01:12):
that's very interesting for yourlisteners if you really believe
them it is.

Speaker 2 (01:17):
This is how I would answer your question.

Speaker 1 (01:19):
Love is an experience of transcendence towards others
, towards the world, towardsnature.
So there's a kind ofself-transcendence that comes
along with a feeling of love, amovement beyond yourself towards
the other.
You're open to awe and wonder.

(01:40):
You're open to a freshappreciation of what would
normally be just seen asfamiliar and mundane and
everyday.

Speaker 2 (01:50):
Hello everyone, welcome to Love and Philosophy.
This is Andrea Hyatt.
Today I'm talking to JulianKiverstein.
He is a senior researcher inthe psychiatry department at
Amsterdam University MedicalCenter.
He writes a lot aboutphenomenology, embodied
cognition, active inference,predictive processing.
You might have heard of hisbook Extended Consciousness and

(02:13):
Predictive Processing, a ThirdWave.
He co-authored that withMichael Kirchhoff.
And today we are going to talkabout play, which is a very
important word for so manyreasons Surprise, uncertainty,
certainty meaning flourishing.
There's a lot of things in thisconversation.
I did it when I was travelingand I'm a bit discombobulated.

(02:33):
If you watch the video you'llsee I'm in sort of a makeshift
kitchen environment andsomething about being off
balance.
I think actually helped us goto some new places.
Took a bit of time.
The first part is going to makepeople happy who want to hear
about Carl Friston and activeinference and predictive
processing, and the second halfwe go into more personal things

(02:55):
or creative ideas.
We talk about a museum calledEl Echo, which I should name
here.
It's a museum in Mexico City.
We'll talk about four paperstowards understanding.
Play its relation to activeinference, which I've name here.
It's a museum in Mexico City.
We'll talk about four paperstowards understanding play, its
relation to active inference,which I've already mentioned.
This conversation with Julian Ihad a while ago it's been maybe
half a year or more and I hadtalked with Carl Friston not too

(03:16):
long before Julian and I talked.
So there's a lot of referencesin here to Carl.
You don't need to listen tothat conversation, but I just
want you to know.
If you hear Carl, that's whowe're talking about.
Also, these kind of big terms,if you haven't heard them before
active inference, free energyprinciple, predictive processing
just think of it as referringto a theory about living systems

(03:36):
which it proposes that we sortof minimize uncertainty about
our sensory inputs by exploring,interacting with the
environment and we have aninternal model of the world, so
to speak, and that can all becontroversial.
This whole internal model thing.
You've probably heard me talkabout quite a lot here, but the
point is just to think of it asthis minimizing surprise,

(03:57):
because actually surprise has alot to do with play, as you'll
hear in this conversation, andhow we deal with uncertainty,
how it develops throughout ourlives, how our lives develop
through play to some extent.
Julian has written quite somepapers about active inference
and he has a different take onit than Carl, so you will hear
us talk about that a little bithere.

(04:17):
He's coming out of more theecological psychology side of
things.
If you remember my conversationwith Harry Heft, we talked
about ecological psychology andthis term affordances, which
doesn't mean they're good or bad, it's just opportunities for
behavior in the environment.
We talked about JJ Gibson, whoHarry Heft wasa student of and

(04:39):
Gibson's one of the founders ofthis and he used this term
affordances.
I'll link to that episode Again.
You don't need to listen to allthose episodes before now, but
there are a lot of threads thatcome from many of the other
discussions I've had.
And I also want to mention theActive Inference Institute which
I've worked with, and there's avideo there between Darius, who

(05:00):
did quite some interestingconversations relative to active
inference with many differentpeople, and he did one with
Julian.
So if you want to go into thereally detailed, nerdy side of
active inference and Julian'swriting about it and
phenomenology, then I willsuggest that conversation and I
will link to it for you.

(05:21):
This conversation is a littlebit lighter.
It definitely gets lighterabout halfway through and we
start really talking about howthese ideas relate to academia,
to our everyday life to where weare right now in a lot of
different ways.
And we do talk about affordances.
So that word just meansopportunities for behavior.
Again, if you think about alake and a rock or an old piece

(05:44):
of wood beside the lake, if Iwalk there it will afford me a
place to sit, it might afford aninsect a home, it might be in
the way of someone who's tryingto put their boat in.
There's all these differentkinds of affordances depending
on who you are, what you are,what your body is, what your
goal is.
So that's all with thataffordances word.

(06:05):
We do talk about them asrelational here.
There's a lot of different waysto think about them.
But this word relational you'veprobably heard come up in a lot
of other podcasts in differentways, maybe with bio, or Karen
Berard has come up and her ideaof relational.
There's a lot of ways ofthinking about it.
This one is more the embodiedcognition and ecological
psychology idea of how we shapeour behavior.

(06:28):
One thing I want to mention hereis that we talk about safety
and this space of safety that isnecessary for play.
I think that's an interestingidea that we think about here
when we talk about playfulness,and that playfulness is actually
a driving force of creation,also of survival.
It's important for us even whenwe're adults.
And what does that mean?

(06:49):
Are we safe to do that now,especially within the realms and
the walls of work and academia?
We talk about surprise and risk, how these are essential for
our development, but also, in aweird, seemingly paradoxical way
, we also need to be safe inorder to really move through
those ideas, those moods, thoseemotions, those experiences that

(07:09):
can allow us surprise and risk.
So how do we have both of thoseat once?
How do we not get stuck in atractor state I think it was Tom
and I talked about attractors abit and Julian and I talk about
it here through more of amedical or psychiatric lens of
these attractors that we sort ofget stuck in, think of it as
sort of looping around andaround in the same patterns or

(07:31):
habits, and sometimes playdisruption, which is a part of
play, can, you know, jolt us outof that, zap us into another
place.
But when it comes to some veryserious illnesses, which we also
talk about here, that looks alittle different, but still it's
worth considering.
It's funny.
I was listening to thisconversation.
I remembered way back, whenGoogle was something fresh and I

(07:57):
don't know.
It was another world and Googleused to have these playgrounds
where you could actually playwhen you work there.
You would have playtime.
It was a new idea, I mean.
I know now workspaces arecompletely different, but they
have almost become somethingelse, not places to play, even
if they're still designed incertain ways for co-working and

(08:18):
so forth.
But I wonder about this idea ofplay, especially in the tech
world, what it really meansthese days and that old spirit
of you know having playgroundsand playtime.
What was that?
It almost seems just really faraway.
Now.
There's a lot of change goingon, obviously, and in that
context this conversation makeseven more sense and fills even

(08:40):
deeper.
Today, in this moment, what doesit mean for us to work together
?
How do we find meaning in thosespaces, those workspaces?
What kind of work do we want todo in the future?
How do we find sensual ways ofbeing alive, helping one another
, getting through?
You'll also hear Julian talkabout how he was really
influenced by another guest thatwas on here, evan Thompson.

(09:01):
So another thread.
Actually there's even more.
But we even get to a questionabout love and how it relates to
all of this.
Surprise, surprise, it comes atthe end.
So today, wherever you are,whatever space you're in, I hope
you find a little time to play,hope you make some space, hope

(09:21):
someone helps you hold thatspace or you find a way to hold
it for yourself, or you findsomeone to help hold the space
for, and just do something thatlifts your spirit.
Feel a little free, free inyour body, as your body in the
world.
Just dance around.
We can all do that.
Find a song you like or youused to like and listen to it

(09:41):
and dance around.
Don't worry if you feel silly,that's part of it.
And if you sit with thesilliness or you let yourself go
into it a little bit, you mightactually find something strange
and delightful happens.
All right, here we go.
Okay, julian, so just to start,how did you get into studying

(10:01):
mind and why play?

Speaker 1 (10:03):
Yeah, so how did I get into doing all of these very
different things about the mindFrom the Varela Thompson and
Roche embodied mind this is whatI discovered as an
undergraduate in the 1990s.
I came across this amazing book, the Embodied Mind, and it was

(10:25):
giving me a way of thinkingabout how to bring together mind
, body and human experience, andas an undergraduate that was
something that was fascinatingme.
I was puzzled by, and here wasa book that managed to tie
together cognitive science,buddhist philosophy,
phenomenology and show how toconnect those very different

(10:49):
traditions to address thisquestion of how to bring
together mind, body andexperience consciousness and how
to fit that into the physicalworld.
And and I was interested inBuddhism as well I hadn't really
encountered phenomenology yet.
And then I read this book andit was bringing together, it was

(11:10):
addressing the questions of howto fit human experience into a
scientific understanding of theworld, but doing that in a way
that also connected to buddhism.
So I found that very inspiring,and all of my work, I should
say, is collaborative.
So I love to work with otherpeople and I find that, uh, I

(11:34):
can write at my best when I'm indialogue with others, and so a
lot of my best papers the onesI'm most proud of are ones that
I've co-authored with otherpeople, and somebody that I've
found very inspiring to workwith is mark miller, who is
based at the university oftoronto, and, uh, with mark
we've been, we've been exploring, like the the dark side of, uh,

(11:58):
mental life.
Uh, by looking at majordepression, addiction and
obsessive-compulsive disorder,with my colleagues here in
Amsterdam I work in a psychiatrydepartment.
So we've been thinking a lotabout how to make sense of
mental illness, the experienceof mental illness and what's

(12:22):
happening neurobiologically.
And, yeah, the predictiveprocessing, free energy ideas as
you're exploring with Carl aswell, actually in relation to
schizophrenia are very helpfulfor providing a framework that
can integrate the biologicalprocesses, the psychological

(12:43):
processes and also the thephenomenological side to mental
illness.
It provides us with formaltools that that really allow us
to bring together theseotherwise difficult to integrate
strands of what is a verycomplex phenomena.
So then we started to think well, could you flip this around and

(13:05):
start to think about how thingscan go well for a person?
So not only thinking about howpredictive processing can lead
to people having what we call asuboptimal grip on the world,
but what would it be forpredictive processing to be able
to model human flourishing orpeople who are happy, people who

(13:31):
feel like they're satisfiedwith their lives.
So that's the big question thatwe had Could you use predictive
processing both to understandmental illness, but also to
understand subjective well-beingor when things go well for a
person in their life?
And yeah, that's what then ledus to think well, maybe there's

(13:54):
a way in which play can help uswith that question of what is it
to be well and for things to gowell.

Speaker 2 (14:01):
And when you were looking in both those cases,
when you're looking at somethingthat we've called mental
illness, what do you thinkyou're doing with the FEP, or
with the free energy principle,or with active inference?
Are you trying to noticepatterns?
Are you trying to find a way toexternally communicate about
what you're noticing in these?
Like, how would you describethat to someone who doesn't

(14:22):
really understand?
Basically, what I'm asking youis what is, what is modeling?

Speaker 1 (14:26):
what do you think of that as there's two ways I want
to answer that question.
The first is to to go back tofrancisco varela again.
He had this idea of creatingwhat he called a circulation
between different ways ofknowing the mind, between a
third person way of knowing themind, which is what we arrive at
through doing science andthrough mathematics, through, in

(14:50):
his case, dynamical systemstheory, which is what he was
using to understand, to formallyunderstand how the brain works,
and then this first personsubjective experience that we
have of being embodied creaturesin the world.
We have this, these differentways of knowing, different
perspectives, and we need, Ithink, the challenge in

(15:12):
cognitive science as I see it,one that you see beautifully
articulated in the embodied mind, is to create this circulation
between these different ways ofknowing.
There's lots to say about thatcirculation, but what the free
energy principle does is give ussome formal tools, I think, for
for um, creating generativepassages, as forela called them,

(15:37):
between our lived experienceand our neurophysiological and
neurobiological processes.
So mathematics gives us thiskind of neutral vocabulary with
which to create this circulationbetween these otherwise
different ways of knowing.

Speaker 2 (15:57):
That's wonderful, and maybe here you should say how
your version or I don't know howto say that, but your take on
free energy and this is a littlebit different or is
illuminating something in a waythat people have found very
helpful.
There's this debate right ofrealism versus instrumentalism
or something like that, where wethink we can either describe
the world or we can predict it,and somehow there's this way

(16:20):
where you can't be both arealist and just think that
theories are tools, and I feellike you've shown, or in your
collaborations, that there'sanother way to think about this,
the literal fallacy.
I think you and your co-writersthe literalist fallacy.

Speaker 1 (16:35):
Yeah, yeah, so maybe you want to for people who are
thinking of active inference.

Speaker 2 (16:39):
In a certain way.
What's important about yourtake on it?
That that you want them to knowyeah, so this is.

Speaker 1 (16:45):
This goes back to the , the model and the map model,
map territory fallacy, how tothink about the relationship
between the free energyprinciple, active inference as a
set of modeling tools, and thesystem, the target system that
we're trying to model, which isthe embodied creature in the

(17:06):
world.
And so I think the first thingI want to say about that is that
it's important to think thatthe model is an abstraction or
an idealization.
And so this goes back to yourconversation with Evan Thompson
recently and the blind spot thatwe shouldn't make the mistake
of taking what is an abstractionand idealization and mistaking

(17:30):
that for reality, and I thinkyou can find places in my
writings where perhaps I'm notsuitably careful about that.
So yeah but I think that thatthat my meta perspective on this
is is always to start fromhuman experience as what is
concretely real, and then weneed to somehow bring that into

(17:55):
conversation and dialogue withwhat we know about our minds
from studying themscientifically.
So there's different ways ofknowing here that we need to
bring into dialogue with eachother.
So what we might do then is givesome kind of literal reality to

(18:16):
the free energy principle as amodel and think, well, if what's
being modeled is real so therereally are these processes that
are where there's predictionerror being computed and there's
predictions that are being madetop down and this is really

(18:37):
going on in the brain then therethere must be this literal
computation taking place in thebrain.
The brain must literally be acomputer, and part of what we
try to do is to say well, that'sone way of being a realist, but
we don't think that realismrequires that kind of literalism

(19:00):
about prediction error,minimization or approximation of
Bayesian inference.
Instead, we can think that themodel is giving us formal tools
and those formal tools aremapping out some reality for us
scientifically.
So that gives us a way of beinga realist about what science is

(19:25):
doing.
But we don't need to think thatthe algorithms that are being
computed here are ones that the,that neurons or the wetware of
the brain is literallyimplementing.
We can think instead that themodels are more like fictional

(19:46):
realities that allow us to docounterfactual modeling, for
instance of the brain.
So they are giving us knowledgeof something real, but we don't
then need to impute them withliteral reality.
These kinds of models.
What we've tried in the in ourphilosophy of science work on

(20:06):
the on the free energy principle.
That gives us like a middle way, I think, between a very
literal understanding of thefree energy principle and a
purely instrumental one, like inhis hippolyto and maxwell
ramstead sometimes offer and inyour discussion with carl as
well, he he tended to use thisformulation of well, it's as if

(20:30):
the brain were doing bayesianinference or were minimizing
free energy and that as ifqualifier.
There is often what encouragespeople to be instrumentalists,
but the philosophy of science,literature, I allows for a more
nuanced position which isneither literalist nor

(20:52):
instrumentalist.
So that's what we've beenexploring.
Michael Pertroff has a new bookon scientific modeling that is
coming out soon with MIT, wherehe develops this line of
argument further.
So he's done a lot morethinking on this since we did
that work.
Yeah, my view is a little bitdifferent from Michael's,
probably that I want to be a bitmore pragmatist.

(21:13):
I think about, about scientificmodels and scientific
explanations.
So the kind of realism that Iwould go for would be more about
keeping the whole context ofthe scientific techniques that
we're using, the phenomena thatare created by the use of those

(21:37):
scientific techniques, and beinga bit more careful about what
realism is exactly, being a bitmore careful about what realism
is exactly, which takes us backto the realism, idealism and the
phenomenological pragmatistideas that are kind of the
foundations for my philosophicalworldview, as it were.

Speaker 2 (21:57):
What's play?
Where did play come in for you?

Speaker 1 (22:03):
How did we end up with play?
That's an excellent question.
It kind of grew out of our workon predictive processing and
active inference.
In that literature you findthis idea of epistemic actions,
so actions that are for seekingout new information, for
resolving uncertainty, andthere's an emotional or affect.