Latest / Mahon McCann / Mark Miller & Ben White: How Social Media Warps your Cognition

Transcript

Mark Miller & Ben White: How Social Media Warps your Cognition · Mahon McCann

0:00
SPEAKER_02: Thank you for joining me, Mark and Ben.
0:02
SPEAKER_02: I was telling you a little bit beforehand that I'm obsessed with your work on
0:07
SPEAKER_02: active inference and social media particularly,
0:09
SPEAKER_02: but also just tech,
0:10
SPEAKER_02: cognition,
0:11
SPEAKER_02: interrelationships.
0:12
SPEAKER_02: So I wondered maybe to start off,
0:14
SPEAKER_02: would you like to say something about that,
0:16
SPEAKER_02: about your collaboration and about your work,
0:19
SPEAKER_02: maybe just to introduce people?
0:21
SPEAKER_00: Oh, that's nice.
0:22
SPEAKER_00: Ben, why don't you take a crack first?
0:24
SPEAKER_00: Because we've got a few things like this now,
0:27
SPEAKER_00: actually,
0:27
SPEAKER_00: you and I.
0:29
SPEAKER_01: Yeah,
0:30
SPEAKER_01: it's been an ongoing collaboration with Mark and myself,
0:34
SPEAKER_01: which has been fantastic.
0:36
SPEAKER_01: I think one of the reasons for that is me and Mark work so well together.
0:41
SPEAKER_00: So well.
0:42
SPEAKER_00: You're one of my favorite collaborators.
0:44
SPEAKER_00: You and Julian Kiverstein.
0:45
SPEAKER_00: I love collaborating with you guys.
0:47
SPEAKER_01: Things happen very easily and kind of organically and the writing is very fluid,
0:53
SPEAKER_01: which is not that common.
0:56
SPEAKER_01: But it started, Mark was actually my master's degree supervisor back at Edinburgh in 2018, 2019.
1:05
SPEAKER_01: And at the time,
1:06
SPEAKER_01: I'd not been exposed to active inference,
1:09
SPEAKER_01: predictive processing,
1:10
SPEAKER_01: or the free energy principle when I started that degree.
1:14
SPEAKER_01: It wasn't until Mark happened to give a seminar at Edinburgh on the topic of the
1:19
SPEAKER_01: free energy principle and predictive processing that I was exposed to those frameworks.
1:23
SPEAKER_01: And so that was definitely a kind of... And is your background in philosophy, Ben, as well?
1:28
SPEAKER_01: It is, yeah.
1:29
SPEAKER_01: So my undergraduate degree was here at Sussex, actually, and it was philosophy and English literature.
1:35
SPEAKER_01: And so originally I was interested in aesthetics, fiction, imagination, all of that kind of stuff.
1:42
SPEAKER_01: And then I kind of found myself.
1:45
SPEAKER_01: It was a very gradual drift over towards philosophy, mind, cognitive science.
1:49
SPEAKER_01: And I was I was fortunate enough to meet one or two people by completely by chance
1:54
SPEAKER_01: along the way that had kind of ongoing influence on me.
1:58
SPEAKER_01: And that culminated with me applying to Edinburgh to their cognitive science
2:02
SPEAKER_01: philosophy of cognitive science program.
2:04
SPEAKER_01: And that's where I've been ever since.
2:07
SPEAKER_01: So I feel like I've kind of found my home and coming into contact with predictive
2:12
SPEAKER_01: processing and active inference was definitely kind of a big part of that because
2:18
SPEAKER_01: maybe we can talk about this later,
2:19
SPEAKER_01: but there are certain things about those frameworks that kind of made me think this
2:23
SPEAKER_01: is where I need to be.
2:25
SPEAKER_01: This is how I'm going to address these questions in the way that I want to address them.
2:28
SPEAKER_01: But my thesis was on social media.
2:30
SPEAKER_01: I knew I wanted to work on technology and Mark supervised that thesis.
2:34
SPEAKER_01: And I think we realized that we just tended to think about these things in the same
2:39
SPEAKER_01: ways and that we worked well together.
2:41
SPEAKER_01: And so we never stopped working together.
2:46
SPEAKER_02: I'm glad you did as well, because it's been very beneficial for myself anyway, reading them.
2:50
SPEAKER_02: I'm going to include the papers and the link for people because I think
2:53
SPEAKER_02: They're very accessible.
2:54
SPEAKER_02: That was something I really found about your work,
2:55
SPEAKER_02: Mark,
2:55
SPEAKER_02: as well,
2:56
SPEAKER_02: as an introduction to active inference and predictive processing,
3:00
SPEAKER_02: that it's really,
3:01
SPEAKER_02: you're one of the best onboarders,
3:02
SPEAKER_02: I think,
3:03
SPEAKER_02: to that sometimes very complicated landscape of ideas.
3:08
SPEAKER_02: And so I've found your work really beneficial in terms of actually getting into it.
3:12
SPEAKER_02: And I know you were a student of Jan Vervaeke's, I believe.
3:15
SPEAKER_00: Yeah.
3:15
SPEAKER_00: Well,
3:16
SPEAKER_00: actually,
3:16
SPEAKER_00: the first one that's relevant is Andy Clark because I did my master's PhD and first
3:22
SPEAKER_00: postdoc with Andy.
3:24
SPEAKER_00: And I was in a postdoc with Andy when I was teaching Ben.
3:28
SPEAKER_00: And then Ben won a place as Andy's doctoral student as well.
3:33
SPEAKER_00: And Andy only takes one doctoral student a year.
3:36
SPEAKER_00: So it's an honor.
3:39
SPEAKER_00: Andy's amazing.
3:41
SPEAKER_00: Anything that I do well, I sort of upward attribute to Andy.
3:45
SPEAKER_00: Andy is the most accessible, coolest philosopher of cognitive science in the world.
3:53
SPEAKER_00: Working with him for almost 10 years leaves its mark,
3:56
SPEAKER_00: I think,
3:57
SPEAKER_00: because he aims a lot for clarity and accuracy.
4:02
SPEAKER_00: So that was good, but yeah, I am working.
4:03
SPEAKER_00: I'm working with John right now.
4:05
SPEAKER_00: I'm at the university of Toronto,
4:06
SPEAKER_00: um,
4:07
SPEAKER_00: part-time and,
4:08
SPEAKER_00: um,
4:09
SPEAKER_00: I'm managing his lab,
4:10
SPEAKER_00: the consciousness and wisdom studies lab.
4:11
SPEAKER_00: So we're collaborating on all sorts of stuff.
4:13
SPEAKER_00: I think you've, have you read our relevance paper, but processing and relevance.
4:17
SPEAKER_02: I have.
4:17
SPEAKER_02: Yeah, definitely.
4:18
SPEAKER_02: I had John on last week.
4:19
SPEAKER_02: Actually, we were doing a conversation.
4:21
SPEAKER_02: I've had a couple of conversations with John and we kind of ongoing stuff around.
4:25
SPEAKER_00: He's lovely.
4:26
SPEAKER_00: Hey, isn't he such a nice person interview.
4:28
SPEAKER_00: he's the best man yeah he's just a legitimately good human being i mean and and and
4:33
SPEAKER_02: brilliant at what he does yeah yeah and i mean it'll probably come into this
4:37
SPEAKER_02: conversation as well because it's you know it's relevant to these topics but i
4:42
SPEAKER_02: suppose to start maybe um in terms of active inference like why do you think this
4:47
SPEAKER_02: is the framework for these tech problems i mean that your work you
4:53
SPEAKER_02: it's something I happen to agree with,
4:55
SPEAKER_02: but like,
4:56
SPEAKER_02: what is it about it that you think captures the kind of social media problem?
5:00
SPEAKER_00: Um, yeah, it's good.
5:01
SPEAKER_00: Can I take a first crack?
5:02
SPEAKER_00: I can have a go.
5:05
SPEAKER_00: No, no, no, no.
5:06
SPEAKER_00: I want to, I want to go first.
5:07
SPEAKER_00: Let me go first.
5:12
SPEAKER_00: Here's what I think.
5:13
SPEAKER_00: This is the first, this is the first reason why I think it's real life.
5:15
SPEAKER_00: Um,
5:17
SPEAKER_00: One of the things that I find very inspiring about thinking of cognitive systems in
5:23
SPEAKER_00: this anticipatory way is that when you get this shift from thinking,
5:29
SPEAKER_00: oh,
5:29
SPEAKER_00: well,
5:29
SPEAKER_00: the world is out there and the nervous system and brain are busy trying to
5:34
SPEAKER_00: interpret that world.
5:36
SPEAKER_00: When you make that shift from that to primarily thinking the brain and the nervous
5:40
SPEAKER_00: system is taking everything it knows about how the world works and how cause and
5:44
SPEAKER_00: effect flow,
5:46
SPEAKER_00: and then it's generating for itself its best estimation about what's happening
5:51
SPEAKER_00: given what it knows and the evidence.
5:54
SPEAKER_00: That shift...
5:57
SPEAKER_00: That shift is not only interesting for the science because it solves a bunch of problems,
6:02
SPEAKER_00: I think,
6:03
SPEAKER_00: and it helps us overcome some puzzles that we've struggled with in perception
6:06
SPEAKER_00: research for a long time.
6:08
SPEAKER_00: But I think it's also very inspiring for the way that we live.
6:11
SPEAKER_00: And I think it's directly relevant for the problems that we're having with
6:15
SPEAKER_00: technology because it means that the world you live in is being constructed by your beliefs,
6:22
SPEAKER_00: by the things you believe.
6:24
SPEAKER_00: So it's no longer the case that you're just like a person who has some beliefs.
6:28
SPEAKER_00: Like, oh, who cares what I'm looking at for eight hours a day?
6:32
SPEAKER_00: They're just beliefs.
6:33
SPEAKER_00: Like, so what?
6:36
SPEAKER_00: Like, no, I'm a whole human.
6:37
SPEAKER_00: The world is out there.
6:38
SPEAKER_00: It's not going to do anything there.
6:41
SPEAKER_00: Like, so what if I have bad belief hygiene where I'm getting inundated with like weird evidence?
6:48
SPEAKER_00: Like, who cares?
6:50
SPEAKER_00: And this flips out to say like, no, you should care.
6:52
SPEAKER_00: This is your reality.
6:54
SPEAKER_00: Like the reality you live in,
6:56
SPEAKER_00: the world you experience and the way that you experience yourself literally is
7:01
SPEAKER_00: constituted by the things that you believe.
7:03
SPEAKER_00: And if you're on a training regimen where your beliefs are subtly drifting.
7:09
SPEAKER_00: either in more extreme ways or pathological ways or just unskillful,
7:13
SPEAKER_00: unhelpful ways,
7:15
SPEAKER_00: you are slowly,
7:16
SPEAKER_00: slowly drifting into a pathological,
7:19
SPEAKER_00: unskillful,
7:20
SPEAKER_00: unhelpful experience of life.
7:22
SPEAKER_00: So I mean,
7:24
SPEAKER_00: if you're only catching this now,
7:27
SPEAKER_00: go and read the paper so you really get this because this isn't only relevant for scientists.
7:32
SPEAKER_00: I think everybody's got to come into contact with this.
7:34
SPEAKER_00: Your reality is built from what you believe.
7:37
SPEAKER_00: That means you have to be really careful about the people you spend time with, what you believe.
7:40
SPEAKER_00: And algorithms don't care about giving you good beliefs.
7:43
SPEAKER_00: They care about giving you the beliefs that get you hot and bothered so you keep coming back.
7:49
SPEAKER_02: Yes,
7:49
SPEAKER_02: I was going to say that there's an entire system around manipulating your beliefs
7:53
SPEAKER_02: to keep you engaged for the profit of companies.
7:56
SPEAKER_02: And the more time we spend in the digital landscape with those incentives,
7:59
SPEAKER_02: I know you,
8:00
SPEAKER_02: from the paper,
8:01
SPEAKER_02: you have this idea that social media is like a bad evidence generator.
8:04
SPEAKER_02: The kind of crowbars are a generative model from reality.
8:10
SPEAKER_02: And,
8:10
SPEAKER_02: you know,
8:11
SPEAKER_02: the more time we spend potentially in the digital world,
8:13
SPEAKER_02: the more disconnected we could get from reality and that there's affective and,
8:18
SPEAKER_02: you know,
8:19
SPEAKER_02: there's so many consequences for that in terms of our own experience of life.
8:23
SPEAKER_02: I don't want to go too far away, Ben.
8:25
SPEAKER_02: If you want to even...
8:26
SPEAKER_02: jump in on that one and then, yeah, maybe this bad evidence generator we can start from there.
8:32
SPEAKER_01: Yeah, so I would certainly just kind of re-emphasize what Mark said.
8:37
SPEAKER_01: I think
8:38
SPEAKER_01: What you get from predictive processing and active inference is this shift towards
8:42
SPEAKER_01: something quite profound,
8:43
SPEAKER_01: which is the notion that what you experience as a subject is not the world.
8:49
SPEAKER_01: You have a generative model that is constituted by sets of expectations,
8:54
SPEAKER_01: priors,
8:55
SPEAKER_01: and precision weightings,
8:57
SPEAKER_01: which is essentially the kind of,
8:59
SPEAKER_01: in case anybody watching is not familiar,
9:02
SPEAKER_01: it's a kind of inference about how reliable certain streams of information are or
9:05
SPEAKER_01: how relevant they are.
9:07
SPEAKER_01: And so what you experience in a subjective way is that model.
9:12
SPEAKER_01: Essentially, you experience the world is kind of constituted by your expectations.
9:16
SPEAKER_01: And so right out of the gate, I think once you adopt this view, it motivates you.
9:25
SPEAKER_01: kind of new deeper concern about things like social media because the things that
9:29
SPEAKER_01: you're taking on the evidence generators that you're exposed to really matter in a
9:35
SPEAKER_01: much deeper way and i think it also shifts us towards maybe even a more
9:40
SPEAKER_01: understanding and compassionate view which is when we see people that are
9:44
SPEAKER_01: struggling with exposure to certain things it gives us a kind of a
9:50
SPEAKER_01: a slightly more striking substantive picture of what's going on there.
9:54
SPEAKER_01: It's not the case that we're just influenced by things that we see.
9:56
SPEAKER_01: Actually, when the conditions are right or wrong,
10:01
SPEAKER_01: that can then come to constitute the way that we experience the world and other
10:04
SPEAKER_01: people in really profound ways.
10:07
SPEAKER_00: So if you think about conspiracy theories and you think just how could somebody be
10:11
SPEAKER_00: on board with that,
10:14
SPEAKER_00: this is part of the answer.
10:16
SPEAKER_00: Just like Ben said, you get the constituent features to be right.
10:20
SPEAKER_00: And then that is the most relevant reality.
10:22
SPEAKER_00: I mean, they are defending, they're defending what they take to be real.
10:27
SPEAKER_00: And that's, that's how powerful, that's how powerful this adjustment can be.
10:30
SPEAKER_00: It's not just biasing, but constituting.
10:32
SPEAKER_00: Yeah.
10:33
SPEAKER_02: And that those beliefs aren't just like a propositional kind of like, Oh, I like these things.
10:37
SPEAKER_02: Like they're very much embodied and it is your reality in a very experiential sense of like,
10:43
SPEAKER_02: you know it it's distorting the way i thought it was very interesting the example
10:47
SPEAKER_02: you give in the paper of uh the snapchat surgery uh like body dysmorphia whereby
10:52
SPEAKER_02: people are actually it's a way of trying to bring together your online self and
10:56
SPEAKER_02: your offline self so you're not registering such a prediction error because that
11:00
SPEAKER_02: seems like almost an absurd thing that you'd like trot out and be like oh look
11:03
SPEAKER_02: these people doing weird stuff and you're like well actually you
11:06
SPEAKER_02: It turns out that's actually a significant way of trying to cope with this
11:09
SPEAKER_02: transformation that's happened that people are just thrown into.
11:12
SPEAKER_01: Yeah, I think it's another important thing.
11:16
SPEAKER_01: I mean, people outside of the predictive processing world have talked about this.
11:21
SPEAKER_01: There's a book by Hugo Mercier and Dan Sperber called The Enigma of Reason that
11:26
SPEAKER_01: talks about the way that our beliefs are
11:29
SPEAKER_01: We take it for granted that our beliefs are just there to be truth tracking in some
11:33
SPEAKER_01: really idealized way.
11:34
SPEAKER_01: But actually,
11:35
SPEAKER_01: our beliefs are the whole process of minimizing prediction error and maintaining
11:41
SPEAKER_01: this kind of optimal grip on your world is sometimes less about truth tracking and
11:46
SPEAKER_01: more about just coping in certain ways.
11:49
SPEAKER_01: So, you know, like the most optimal model is not always going to be the most truth tracking model.
11:55
SPEAKER_01: which is something else that comes out of predictive processing.
11:58
SPEAKER_01: But another thing as well is one thing that seems really unique to these frameworks is
12:05
SPEAKER_01: So I kind of came into PP through 4E cognitive science,
12:08
SPEAKER_01: which has this emphasis on embodiment and extension and embeddedness within an environment.
12:14
SPEAKER_01: And I think there's a sense in which there's an irony in the recent history of
12:18
SPEAKER_01: cognitive science in that all of these frameworks and approaches were kind of
12:21
SPEAKER_01: pushing outwards the boundaries.
12:23
SPEAKER_01: So it was trying to get beyond the brain in a way and into the significance of cognitive
12:28
SPEAKER_01: how we act on the world and how the way that we think and feel is kind of deeply
12:37
SPEAKER_01: constituted and influenced and scaffolded by external things.
12:41
SPEAKER_01: But there was a kind of emptiness at the core of this,
12:44
SPEAKER_01: which was that there was,
12:45
SPEAKER_01: and this is something Andy has said,
12:46
SPEAKER_01: which is there was never really any kind of cohesive theory of brain function at
12:50
SPEAKER_01: the center of this.
12:51
SPEAKER_01: So we had this pushing outwards, but there was emptiness.
12:53
SPEAKER_01: There was nothing, there was no account of what was going on on the inside.
12:57
SPEAKER_01: And that's what active inference has given us.
12:59
SPEAKER_01: So now we have this very complete picture because active inference is this theory
13:04
SPEAKER_01: of brain function that has these formalisms and this quite sophisticated
13:08
SPEAKER_01: mathematical side,
13:09
SPEAKER_01: but it can also accommodate these other insights from embodied cognitive science.
13:13
SPEAKER_01: And so what you end up with is if you tell the story in a very comprehensive way,
13:18
SPEAKER_01: you end up with a very multi-layered
13:20
SPEAKER_01: um, notion of what an agent in the world is.
13:24
SPEAKER_01: So you can move from neurobiology,
13:27
SPEAKER_01: uh,
13:27
SPEAKER_01: computation through to like the importance of introception and physiology and
13:32
SPEAKER_01: embodiment in this very raw,
13:33
SPEAKER_01: very visceral way,
13:34
SPEAKER_01: plus the significance of action,
13:37
SPEAKER_01: which,
13:37
SPEAKER_01: which is,
13:37
SPEAKER_01: you know,
13:38
SPEAKER_01: a core part of active inference.
13:40
SPEAKER_01: And,
13:41
SPEAKER_01: um,
13:41
SPEAKER_01: Andy's work as well has shown how all of this accommodates,
13:44
SPEAKER_01: um,
13:44
SPEAKER_01: not just accommodates,
13:45
SPEAKER_01: but actually kind of
13:47
SPEAKER_01: really makes sense of and augments ideas about the extended mind and embedded cognition,
13:55
SPEAKER_01: distributed cognition.
13:56
SPEAKER_01: So it just kind of brings everything together in a really nice way.
13:58
SPEAKER_01: And you end up,
14:00
SPEAKER_01: and this is something I really like about Mark's work on addiction,
14:03
SPEAKER_01: is that you end up with this kind of agent environment system account of a phenomenon.
14:08
SPEAKER_01: It was exactly what I was thinking.
14:09
SPEAKER_01: There is just an account of like what's going on in the dopamine or whatever.
14:13
SPEAKER_01: what's happening outside like it actually connects all of these things in a really
14:17
SPEAKER_02: nice way and it seems essential for social media as well because the digital
14:21
SPEAKER_02: environment is an environment that we're in and for me a lot of the attention
14:25
SPEAKER_02: economy literature was very much missing this kind of framework of environment
14:30
SPEAKER_02: agent dynamics like they might have had kind of a thin account of the agent but
14:34
SPEAKER_02: then there was a sense that it was just this perceptual thing going on it's like no
14:38
SPEAKER_02: actually
14:39
SPEAKER_02: embedded in a digital environment like an extended cognition kind of perspective of
14:44
SPEAKER_02: like your brain is taking these things as real like you know there isn't a huge
14:48
SPEAKER_02: difference
14:49
SPEAKER_02: Between the experiences online and the experiences offline.
14:51
SPEAKER_01: One of the things I really like about Yele's work on this is this kind of shift
14:55
SPEAKER_01: from seeing the problem as just an abundance of information.
14:58
SPEAKER_01: So, you know, the mind is just a kind of passive mind.
15:00
SPEAKER_02: Yes, that's literally what I'm writing about at the moment.
15:02
SPEAKER_01: Taking in information where Yele uses active inference to say,
15:05
SPEAKER_01: no,
15:06
SPEAKER_01: look,
15:06
SPEAKER_01: it's a little bit more complicated than that.
15:08
SPEAKER_01: And it's about the affordances and the kind of the opportunities to act that these
15:13
SPEAKER_01: digital affordances are giving us.
15:14
SPEAKER_01: I think that was, he articulates that really nicely with active inference.
15:19
SPEAKER_02: And it's a massive change really in the literature because the attention economy
15:22
SPEAKER_02: literature is based on that filter model of attention,
15:25
SPEAKER_02: really the kind of broad Bentian paradigm of this like serial processing of just
15:29
SPEAKER_02: information coming in and then action being taken after that.
15:33
SPEAKER_02: Whereas whenever the Yele reverses it in terms of active inference, you get a very different...
15:39
SPEAKER_02: picture really of why people would be so immersed in this and then you get like the
15:42
SPEAKER_02: social competition dynamics of like you know like getting metrics for popularity
15:48
SPEAKER_02: and how important you are and social competition for attention and you almost get
15:53
SPEAKER_02: another feedback loop that keeps people involved beyond just the like skinner box
15:57
SPEAKER_02: aspect of like rewards for this and that um so that's my framing of
16:03
SPEAKER_02: Yeah, sorry.
16:04
SPEAKER_02: Maybe I'm going too quickly there.
16:05
SPEAKER_00: No, that's good.
16:06
SPEAKER_00: That's good.
16:06
SPEAKER_00: I just wanted to put a tail on that.
16:07
SPEAKER_00: I mean,
16:08
SPEAKER_00: we've been talking about the way that social media can harmfully bend your model of
16:12
SPEAKER_00: how the world works.
16:13
SPEAKER_00: And you just touched on such an important part there.
16:16
SPEAKER_00: One of the places that it can bend us is it can bend your expectations for the
16:19
SPEAKER_00: kinds of rewards you're expecting in a particular niche,
16:22
SPEAKER_00: which really turns out to be very vicious.
16:25
SPEAKER_00: So, you know, we're set up to track and especially feel good when things are working in a certain way.
16:34
SPEAKER_00: We have a story about that.
16:35
SPEAKER_00: I mean, we all know that, right?
16:36
SPEAKER_00: Like when you eat and you drink and when we sex,
16:40
SPEAKER_00: when we,
16:40
SPEAKER_00: you know,
16:41
SPEAKER_00: these are things that we just have hardwired rewards for.
16:44
SPEAKER_00: And we have created these synthetic digital replacements for a lot of these things.
16:52
SPEAKER_00: So you're getting cued that you're getting this huge amount of social credit,
16:58
SPEAKER_00: which feels incredible for the system that's attempting to produce high social credit.
17:04
SPEAKER_00: But two things happen quick.
17:06
SPEAKER_00: One, it's empty social credit.
17:08
SPEAKER_00: I like saying this to my students.
17:12
SPEAKER_00: First of all, I make the game.
17:14
SPEAKER_00: who has the most followers and likes in the room.
17:17
SPEAKER_00: We sort of suss it out until somebody says, you know, I've got a million, like I've got a super cute cat.
17:23
SPEAKER_00: So I've got like a million.
17:24
SPEAKER_00: And then you say,
17:26
SPEAKER_00: how many people on that list are coming when you're in hospice to help change your diaper,
17:32
SPEAKER_00: to feed you?
17:33
SPEAKER_00: How many on that list?
17:35
SPEAKER_00: How many on that list when you come to end of life are coming to be with you?
17:41
SPEAKER_00: Because that's actually a really good way to think of how many good friends you have.
17:44
SPEAKER_00: How many friends are going to show up when it gets messy and ugly?
17:47
SPEAKER_00: And the dangerous thing here is that we evolved the sensitivity for social credit.
17:54
SPEAKER_00: So that we were motivated to create networks and dynamics of support systems and so
18:02
SPEAKER_00: that we can reduce this volatility,
18:03
SPEAKER_00: this entropy together as a community.
18:05
SPEAKER_00: That's the whole thing.
18:06
SPEAKER_00: One of the great things about humans is we're like Voltron.
18:08
SPEAKER_00: We can combine to become better at managing uncertainty.
18:13
SPEAKER_00: We're better together, you know, and that's what this is.
18:15
SPEAKER_00: And it's been completely hijacked because you can give you the sense that it's
18:22
SPEAKER_00: working when in fact it's completely hollow.
18:24
SPEAKER_00: It's just like heroin in a way,
18:26
SPEAKER_00: because heroin makes the brain feel like everything you've ever wanted,
18:30
SPEAKER_00: you're doing better than expected at getting it now,
18:32
SPEAKER_00: except for actually the opposite's happening is you're actually trending away,
18:36
SPEAKER_00: trending away from a good grip rather than towards a good grip.
18:38
SPEAKER_00: so nasty actually.
18:40
SPEAKER_00: And we think it's so innocuous, but it isn't.
18:42
SPEAKER_00: It's got a really nasty side effect.
18:44
SPEAKER_00: And once you have those expectations set, anything less than that burns us.
18:49
SPEAKER_00: So like once you get onto heroin is the amount of air that you can reduce,
18:53
SPEAKER_00: then the brain is like,
18:54
SPEAKER_00: well,
18:54
SPEAKER_00: heroin is the only thing I want to do because nothing else satisfies like that
18:58
SPEAKER_00: anymore because I have expectations that that's how good we can feel.
19:01
SPEAKER_00: Same thing with social credit.
19:03
SPEAKER_00: You feel like,
19:03
SPEAKER_00: oh,
19:03
SPEAKER_00: now anything less than 10,000 likes and I don't like that picture and I don't like
19:08
SPEAKER_00: myself in that picture,
19:10
SPEAKER_00: you know,
19:11
SPEAKER_00: or I don't like myself at all because my online filtered image gets all this
19:15
SPEAKER_00: attention and my real life image doesn't like who I actually am doesn't.
19:18
SPEAKER_00: So I don't like myself anymore because there's the social credit I'm used to now.
19:22
SPEAKER_00: You're continually chasing the dragon.
19:26
SPEAKER_02: Yeah, and that it's leapt over the physical world in a sense.
19:29
SPEAKER_02: I really think since COVID,
19:30
SPEAKER_02: my perception was that the digital world started to become more important for a lot
19:34
SPEAKER_02: of people,
19:34
SPEAKER_02: even for news.
19:35
SPEAKER_02: And as this kind of portal into reality,
19:39
SPEAKER_02: as more people got status and fame through social media,
19:42
SPEAKER_02: and that became like the media outlets have to be on social media,
19:45
SPEAKER_02: the institutions have to be on social media.
19:47
SPEAKER_02: the police the guards in ireland actually have to have a like a ministry of social
19:51
SPEAKER_02: media basically where they have to correct in real time the stories that are coming
19:54
SPEAKER_02: out to be like no this isn't true this isn't happening or this is like so um i
19:59
SPEAKER_02: think yeah so many of those dynamics now have become culture really i mean talking
20:03
SPEAKER_02: about distributed cognition like the digital started out as a fun toy thing but i
20:08
SPEAKER_02: think they're really powering a lot of our um international culture
20:14
SPEAKER_01: I think digital technology is interesting in so many different ways,
20:18
SPEAKER_01: but I think just to take a step back,
20:20
SPEAKER_01: I think...
20:22
SPEAKER_01: one of the things that is so dangerous also, I always, I always want to say potentially dangerous.
20:28
SPEAKER_01: I mean, lots of, lots of people I know, uh, engage with social media and completely non-problematic ways.
20:34
SPEAKER_01: And it's,
20:34
SPEAKER_01: so the claim here,
20:35
SPEAKER_01: I always feel,
20:37
SPEAKER_01: I always feel like I need to be careful and say,
20:39
SPEAKER_01: you know,
20:40
SPEAKER_01: the claim is not that social media is like evil.
20:42
SPEAKER_01: It's always deadly in these ways.
20:44
SPEAKER_01: Um,
20:45
SPEAKER_01: I think the real key for this work is to try and identify the kinds of contexts and
20:49
SPEAKER_01: factors that can make it disrupt our dynamic with the world in these ways that can
20:55
SPEAKER_01: be suboptimal or harmful.
20:57
SPEAKER_01: But technology in general, I think, one of the things that it's done is imbued us with this
21:03
SPEAKER_01: incredible capacity to abstract,
21:05
SPEAKER_01: to take these rewards that occur naturally in the world and then abstract them and
21:10
SPEAKER_01: manufacture them in ways that kind of accentuates and blows up their rewarding
21:18
SPEAKER_01: qualities in ways that can really warp our relationship with the world.
21:23
SPEAKER_01: So
21:24
SPEAKER_01: There's a book called Supernormal Stimulus where there's an experiment.
21:30
SPEAKER_01: I forget who did the experiment,
21:31
SPEAKER_01: but it's basically what they did was they took a certain species of bird that lays
21:36
SPEAKER_01: eggs with a particular pattern on.
21:38
SPEAKER_01: And the researchers made fake eggs where the pattern had been kind of made more
21:43
SPEAKER_01: colorful and brighter and bigger.
21:46
SPEAKER_01: and they put the fake eggs into the environment with some parent birds,
21:50
SPEAKER_01: and the parent birds actually abandon their own eggs to sit on the fake eggs.
21:55
SPEAKER_01: And this seems to be something that we're really good at replicating in our world
22:00
SPEAKER_01: with things like sugary junk food,
22:03
SPEAKER_01: with the kind of social interactions that you get on social media,
22:07
SPEAKER_01: with internet pornography,
22:08
SPEAKER_01: I think is a slightly controversial example,
22:10
SPEAKER_01: but I think it does fit this kind of schematic there,
22:14
SPEAKER_01: where...
22:15
SPEAKER_01: The affordances of the technology allow us to kind of ostensibly access more of a reward,
22:23
SPEAKER_01: but it's in the abstraction that we lose something,
22:27
SPEAKER_01: which is what Mark was talking about there.
22:28
SPEAKER_01: I think there's always a nourishing side to these rewards that I think gets lost in
22:34
SPEAKER_01: the abstracting,
22:35
SPEAKER_01: which is really easy to forget.
22:38
SPEAKER_01: And the affordances of digital technology in particular, I think, are really good at doing this.
22:44
SPEAKER_02: Yeah.
22:45
SPEAKER_02: Yeah, and that it's kind of baked into the mechanics of the platforms in some sense.
22:48
SPEAKER_02: So I think people don't notice it necessarily while they're operating.
22:51
SPEAKER_02: Like Mark, you mentioned uploading a photo.
22:53
SPEAKER_02: You think I'm uploading this photo because I like it or something.
22:56
SPEAKER_02: Maybe then you're kind of thinking, oh, what will people think of it?
22:58
SPEAKER_02: And then you're thinking about other things you've posted.
23:01
SPEAKER_02: And so there's a lot of tailoring to like the game mechanics of the environment that goes on.
23:06
SPEAKER_02: And I suppose that one question I wanted to ask you is like,
23:09
SPEAKER_02: do you think it's social media is reformable,
23:12
SPEAKER_02: like in a sense that you could remove some of the persuasive design elements,
23:17
SPEAKER_02: maybe some of the gamification and have a healthier version of it?
23:20
SPEAKER_02: Or do you think it's just doomed?
23:23
SPEAKER_00: I mean,
23:23
SPEAKER_00: it seems like it's possible,
23:25
SPEAKER_00: yeah,
23:26
SPEAKER_00: but I don't know if we won't be attracted to it in the same way.
23:31
SPEAKER_02: No money in it,
23:33
SPEAKER_00: but...
23:34
SPEAKER_00: Yeah,
23:34
SPEAKER_00: you wouldn't scroll for eight hours something that wasn't high sugar,
23:41
SPEAKER_00: high fat.
23:42
SPEAKER_00: In these sorts of ways, probably.
23:45
SPEAKER_00: You would come there and you would, you know, satisfy whatever that tool was meant for.
23:51
SPEAKER_00: And then you would return once again to your,
23:55
SPEAKER_00: you know,
23:55
SPEAKER_00: semi-bumpy,
23:58
SPEAKER_00: commonly socially awkward human existence,
24:01
SPEAKER_00: which is actually...
24:02
SPEAKER_00: Beautiful.
24:03
SPEAKER_00: I mean,
24:04
SPEAKER_00: you don't want to miss you don't want to miss your socially awkward,
24:07
SPEAKER_00: slightly weird,
24:08
SPEAKER_00: bumpy human existence in favor of something polished.
24:11
SPEAKER_00: And because because this is actually what it is to be a human is to be socially
24:17
SPEAKER_00: awkward a little bit and to work through those things and to figure each other out and.
24:21
SPEAKER_00: There's another of those nasty loops.
24:23
SPEAKER_00: Strange loops are everywhere in this framework.
24:28
SPEAKER_00: Everything seems like it eats itself once it gets going.
24:31
SPEAKER_00: But there's another one.
24:33
SPEAKER_00: The more time that we spend online because it feels good and ordinary human
24:38
SPEAKER_00: interaction can be a bit bumpy and a bit awkward.
24:42
SPEAKER_00: So it's smoother and cooler.
24:43
SPEAKER_00: Like real life sex can be bumpy and awkward.
24:47
SPEAKER_00: Online sex is smooth and polished.
24:50
SPEAKER_00: Of course, it's not your sex online.
24:53
SPEAKER_00: So that's, you know, significant.
24:57
SPEAKER_00: But the more time you're in the polished realm,
24:59
SPEAKER_00: the less time you're actually developing the skills to navigate the bumpy,
25:04
SPEAKER_00: awkward,
25:04
SPEAKER_00: but very real human experience.
25:07
SPEAKER_00: And so the less capable you are at it,
25:09
SPEAKER_00: which means the more likely you are to go back online and which deviates from the
25:15
SPEAKER_00: time that you need to actually get better at being human.
25:17
SPEAKER_00: I mean, and once you're in, I mean,
25:19
SPEAKER_00: To get out again,
25:20
SPEAKER_00: I guess I wouldn't be surprised that we find in the research,
25:24
SPEAKER_00: I haven't seen any of this yet,
25:25
SPEAKER_00: but I wouldn't be surprised if it's already out there.
25:27
SPEAKER_00: I mean, I guess it's probably already out there.
25:29
SPEAKER_00: But,
25:29
SPEAKER_00: you know,
25:30
SPEAKER_00: they talk today about different drugs of addiction actually have a natural shelf
25:34
SPEAKER_00: life for most people.
25:37
SPEAKER_00: Like there are like a certain number of years per drug.
25:40
SPEAKER_00: Basically,
25:41
SPEAKER_00: anybody who gets bad into that drug ends up popping out the other side after a
25:45
SPEAKER_00: certain amount of time.
25:46
SPEAKER_00: I suspect,
25:47
SPEAKER_00: I haven't done research on this yet,
25:48
SPEAKER_00: but I suspect what's happening there is that's about the amount of time that you get,
25:53
SPEAKER_00: that you're in a drug before your brain gets enough counter evidence for the way
25:58
SPEAKER_00: that it's communicating that things are going well.
26:01
SPEAKER_00: So like heroin says, no, things are going really well.
26:03
SPEAKER_00: And you need about 10 years of counter evidence and like the harmful effects of a
26:07
SPEAKER_00: 10 year addiction before you're like,
26:08
SPEAKER_00: oh no,
26:09
SPEAKER_00: that's just a feeling.
26:10
SPEAKER_00: Actually, nothing's getting better.
26:12
SPEAKER_00: It's all getting worse.
26:13
SPEAKER_00: Yeah.
26:14
SPEAKER_00: I wonder if we won't do the same thing with social media where we get a natural,
26:17
SPEAKER_00: we get a natural eviction.
26:19
SPEAKER_00: Um, we just aren't at that horizon yet, you know, cause we're on there.
26:23
SPEAKER_00: You do it long enough.
26:24
SPEAKER_00: Then you realize you look around and go like, wait, but I don't have a family.
26:27
SPEAKER_00: I don't have a family or friends or anybody who cares about me.
26:30
SPEAKER_00: So I mean, you have to pop back out eventually.
26:33
SPEAKER_01: Yeah.
26:33
SPEAKER_01: There are always other competing tendencies in life,
26:36
SPEAKER_01: especially,
26:37
SPEAKER_01: I mean,
26:38
SPEAKER_01: I think people who become badly addicted to drugs,
26:42
SPEAKER_01: uh,
26:44
SPEAKER_01: I think those are, in some ways, very egregious cases.
26:48
SPEAKER_01: And I think that you have cases of drug and alcohol addiction where the dynamics really break down badly.
26:54
SPEAKER_01: But I think for most people who struggle with their online life and their social
26:59
SPEAKER_01: media interactions,
27:00
SPEAKER_01: I think
27:01
SPEAKER_01: typically there are going to be competing tendencies there,
27:03
SPEAKER_01: like the desire for good relationships,
27:05
SPEAKER_01: the desire for good friendships,
27:06
SPEAKER_01: the desire to kind of go and see the world and travel and things.
27:09
SPEAKER_01: So I think you,
27:11
SPEAKER_01: this is why it,
27:12
SPEAKER_01: you know,
27:12
SPEAKER_01: I think it's the dynamics and the underlying kind of like the underlying
27:19
SPEAKER_01: foundations that are the same.
27:20
SPEAKER_01: I don't,
27:20
SPEAKER_01: I don't think there's any difference in kind between social media addiction and
27:25
SPEAKER_01: heroin addiction.
27:25
SPEAKER_01: I think it's important observation as well.
27:29
SPEAKER_01: Yeah.
27:30
SPEAKER_01: it's a difference in kind of how badly things have gone wrong and obviously there
27:34
SPEAKER_01: are reasons for that but in terms of
27:37
SPEAKER_01: kind of,
27:38
SPEAKER_01: there's no reason,
27:39
SPEAKER_01: there's no,
27:40
SPEAKER_01: I don't think there's any principled reason that we can't imagine social media platforms.
27:43
SPEAKER_01: In fact,
27:44
SPEAKER_01: I think it's quite easy to imagine social media platforms where you kind of strip
27:48
SPEAKER_01: out some of these more sticky,
27:52
SPEAKER_01: more potentially problematic features.
27:54
SPEAKER_01: I think it's quite easy to look at social media and say what's wrong with it.
27:58
SPEAKER_01: And lots of people have been doing that.
28:02
SPEAKER_01: However, I agree with Mark in that
28:06
SPEAKER_01: Firstly, there's a kind of economic side to this.
28:08
SPEAKER_01: You have to think about what kinds of incentives are in play and what kind of
28:11
SPEAKER_01: business models make sense for people that make social media.
28:15
SPEAKER_01: But also there's a kind of more fundamental issue,
28:18
SPEAKER_01: which is I heard it said recently,
28:21
SPEAKER_01: this was said by Axel Constant,
28:22
SPEAKER_01: who is an active inference researcher.
28:24
SPEAKER_01: So credit to Axel for this.
28:25
SPEAKER_01: But he said that there's something...
28:29
SPEAKER_01: law-like about uncertainty.
28:30
SPEAKER_01: It's kind of like gravity in that any system that is a living system kind of has to minimize uncertainty.
28:38
SPEAKER_01: It's always kind of moving down a gradient of uncertainty towards certainty.
28:43
SPEAKER_01: And so if you think about it in that way, there's
28:45
SPEAKER_01: what we've done in designing digital technology is kind of create this portal for
28:51
SPEAKER_01: uncertainty minimization for doing potentially doing much better than expected.
28:56
SPEAKER_01: Uncertainty.
28:56
SPEAKER_01: And so the question is, are we capable?
29:01
SPEAKER_02: You froze there for a second, Ben, sorry.
29:04
SPEAKER_02: You froze after saying the slope of uncertainty minimizing uncertainty.
29:10
SPEAKER_01: Oh, okay.
29:12
SPEAKER_01: So for any system that's a living system,
29:15
SPEAKER_01: there's this kind of law-like nature of a gradient from uncertainty to certainty.
29:20
SPEAKER_01: And so in the environment,
29:22
SPEAKER_01: we tend to find different strategies and different ways of minimizing uncertainty.
29:26
SPEAKER_01: We have different action policies.
29:28
SPEAKER_01: And in designing social media, I think what we've done is kind of open up this portal
29:33
SPEAKER_01: to very very steep gradients of uncertainty minimization kind of within our within
29:39
SPEAKER_01: the context of our social lives and so i would i i'm interested to think about um
29:46
SPEAKER_01: the possibilities for like what motivations would would we have really for moving
29:51
SPEAKER_01: away from that these things are so so stimulating and so arousing and um
29:58
SPEAKER_01: They can also be portals to very steep gradients of uncertainty minimization in a good way as well.
30:03
SPEAKER_01: So again, it's not all bad here.
30:05
SPEAKER_01: So it's how do you separate out the good from the bad in terms of how effective
30:11
SPEAKER_01: they are at minimizing uncertainty.
30:13
SPEAKER_01: So I think in principle, it's a very easy question.
30:16
SPEAKER_01: We can just imagine...
30:18
SPEAKER_01: We can engage our radical imagination and say,
30:20
SPEAKER_01: oh,
30:20
SPEAKER_01: well,
30:21
SPEAKER_01: we remove the gamification,
30:22
SPEAKER_01: we ban the filters,
30:23
SPEAKER_01: we do this,
30:24
SPEAKER_01: we do that.
30:24
SPEAKER_01: But actually,
30:25
SPEAKER_01: I think in practice,
30:26
SPEAKER_01: I'd be quite skeptical of these platforms flourishing,
30:31
SPEAKER_01: you know,
30:32
SPEAKER_01: for a variety of reasons.
30:33
SPEAKER_01: I don't think we'd be as drawn to use them.
30:35
SPEAKER_01: And I think companies would be less inclined, less incentivized to build them as well.
30:41
SPEAKER_02: Yeah, and I totally agree.
30:42
SPEAKER_02: And I think the business model, which is a lot of what my work is on, basically denies that.
30:48
SPEAKER_02: It's impossible for them really to capture that much attention without having a lot of these features.
30:53
SPEAKER_02: And if one of them does it, if TikTok optimizes for more and more short-form video,
30:58
SPEAKER_02: YouTube has to do it and Instagram has to do it because that's capturing more attention.
31:01
SPEAKER_02: So you end up with this kind of race for whoever figures out the new thing, everybody has to do it.
31:06
SPEAKER_02: And I think in that sense that from within the company's reform is very,
31:10
SPEAKER_02: very unlikely that they do it themselves.
31:13
SPEAKER_02: I think it has to be some sort of regulatory, like public health, like,
31:17
SPEAKER_02: this is you know it's like smoking again basically i've been like these behavioral
31:21
SPEAKER_02: addictions are causing destruction to individuals our culture is becoming addicted
31:25
SPEAKER_02: and it's actually very hard to you know plan for the future and do things with all
31:30
SPEAKER_02: the stuff going on and it'll have to be then the problem is it's an international
31:34
SPEAKER_02: problem so which government you know does it and that stuff but
31:36
SPEAKER_02: I think you could imagine like a better one down the line.
31:40
SPEAKER_00: Part of this complexity is what leads me back to thinking the solution.
31:45
SPEAKER_00: I mean, I'm happy people are working at that sort of solving it out there set, which I guess is needed.
31:51
SPEAKER_00: But I really think part of the solution is going to have to be you solve it in here.
31:54
SPEAKER_00: I think with these sort of twin factors of we don't have good skilled attention abilities,
32:03
SPEAKER_00: these can be trained.
32:05
SPEAKER_00: There are training regimens that are really good at increasing your metacognition,
32:09
SPEAKER_00: your controllability of attention.
32:11
SPEAKER_00: Also,
32:12
SPEAKER_00: there are techniques and trainings and skills that you can generate that allow you
32:16
SPEAKER_00: to access and get more from rewards that are available in your environment.
32:20
SPEAKER_00: And then the third one is,
32:22
SPEAKER_00: you know,
32:23
SPEAKER_00: starting to take bites out of this meaning crisis we're in because one of the
32:28
SPEAKER_00: reasons we are fascinated
32:30
SPEAKER_00: by these better than expected gradients is because in our life,
32:35
SPEAKER_00: given the kind of belief networks that we have and the kind of niches we're
32:38
SPEAKER_00: embedded in,
32:39
SPEAKER_00: we are up against tons of volatility.
32:41
SPEAKER_00: So we're turning to this to get relief.
32:43
SPEAKER_00: We're not necessarily turning to this because it's great.
32:46
SPEAKER_00: We're turning to it because it's better than the thing we're going through.
32:49
SPEAKER_00: And there are ways of changing your relationship.
32:53
SPEAKER_00: Sorry, tell me.
32:55
SPEAKER_02: Better than the void of meaning that people are finding and
32:58
SPEAKER_02: The flow you talk about,
32:59
SPEAKER_02: like the bad bootstraps,
33:00
SPEAKER_02: like flow from video games,
33:02
SPEAKER_02: flow from social media,
33:03
SPEAKER_02: can feel better than a complete absence of it.
33:06
SPEAKER_00: Completely.
33:07
SPEAKER_00: But then maybe the thing to do isn't more regulation.
33:10
SPEAKER_00: I don't want to say that.
33:10
SPEAKER_00: I'm happy people are on that.
33:13
SPEAKER_00: But for me,
33:14
SPEAKER_00: the place that I feel positioned is thinking more about how do we change ourselves
33:19
SPEAKER_00: and how do we change our microcultures.
33:21
SPEAKER_00: Yeah.
33:22
SPEAKER_00: So that we can be extracting more rewards so that we can have better control of our
33:26
SPEAKER_00: attention so that we can have more meaning.
33:28
SPEAKER_00: And,
33:28
SPEAKER_00: you know,
33:29
SPEAKER_00: ideally you would leave these things as a side effect rather than,
33:32
SPEAKER_00: you know,
33:34
SPEAKER_00: trying to prevent the addictive behavior,
33:37
SPEAKER_00: fill it in.
33:37
SPEAKER_00: I mean, we have this from rat experiments.
33:39
SPEAKER_00: I think everybody knows, right?
33:41
SPEAKER_00: Yeah.
33:41
SPEAKER_00: Do you know the rat experiment, man?
33:43
SPEAKER_00: Do you know this?
33:44
SPEAKER_02: Oh, like getting them addicted to cocaine?
33:47
SPEAKER_02: The rat city.
33:49
SPEAKER_02: Good friends and living a good life.
33:53
SPEAKER_00: Exactly.
33:53
SPEAKER_00: As soon as you put them in the rat park where they have friends and complexity,
33:57
SPEAKER_00: then they don't go to the drugs anymore because they can satisfy their...
34:02
SPEAKER_00: They can satisfy their reduction of uncertainty in all sorts of meaningful and interesting ways.
34:06
SPEAKER_00: So they don't need to do it synthetically anymore.
34:08
SPEAKER_00: One of the reasons we might be going to synthetic is because we're not learning the
34:12
SPEAKER_00: skills to have a meaningful life.
34:13
SPEAKER_00: Those are skills.
34:15
SPEAKER_00: There's the good news of the podcast, right?
34:16
SPEAKER_00: I mean, there are actually skills you can develop that make you better at this.
34:20
SPEAKER_01: I think niche construction was mentioned there.
34:22
SPEAKER_01: I think that's really key.
34:23
SPEAKER_01: I think oftentimes in predictive processing,
34:26
SPEAKER_01: you'll hear people say things like,
34:29
SPEAKER_01: oh,
34:29
SPEAKER_01: there are two ways to reduce uncertainty.
34:32
SPEAKER_01: There's update your model or act on the world.
34:36
SPEAKER_01: But there's kind of a third way,
34:38
SPEAKER_01: which is one of the things I like about predictive processing is it does kind of
34:43
SPEAKER_01: bring together the inner and the outer.
34:45
SPEAKER_01: And I think that...
34:48
SPEAKER_01: one thing that we know is really effective is to alter your environment in some way,
34:53
SPEAKER_01: such that you scaffold kind of effective competing tendencies.
34:58
SPEAKER_01: So if you think about your engagement,
35:01
SPEAKER_01: I've done some work on this with Inés Cipollito,
35:03
SPEAKER_01: where we kind of bring in some of the concepts from dynamical systems theory,
35:08
SPEAKER_01: such as like attractor basins.
35:10
SPEAKER_01: And you think about kind of, you think about yourself as like a,
35:14
SPEAKER_01: kind of agent going through life along a trajectory and your social media use is
35:20
SPEAKER_01: kind of getting out of control you're maybe using it more than you wanted to you
35:23
SPEAKER_01: think of that as a kind of basin that with these steep sides that you've been kind
35:27
SPEAKER_01: of sucked into and conceptually one thing i really like about this is you you can
35:31
SPEAKER_01: imagine kind of the remedy to that to just be to flatten those
35:36
SPEAKER_01: Um,
35:36
SPEAKER_01: kind of like the,
35:37
SPEAKER_01: the,
35:37
SPEAKER_01: the size of that attractor basin and some ways you can do that is I think,
35:42
SPEAKER_01: um,
35:42
SPEAKER_01: you know,
35:43
SPEAKER_01: like,
35:43
SPEAKER_01: like Mark said,
35:44
SPEAKER_01: I mean,
35:44
SPEAKER_01: you can focus on your,
35:46
SPEAKER_01: your outside social life.
35:47
SPEAKER_01: You can focus on building different kinds of connections,
35:49
SPEAKER_01: like creative,
35:51
SPEAKER_01: creative hobbies and outlets and all kinds of things.
35:54
SPEAKER_01: Really.
35:55
SPEAKER_01: You're not, um,
35:56
SPEAKER_01: not at all powerless.
36:00
SPEAKER_01: So I think part of the solution is in cultivating competing tendencies in terms of
36:09
SPEAKER_01: the whole network of the things that you're doing.
36:11
SPEAKER_01: You can flatten the steepness of that base in that way.
36:14
SPEAKER_01: But also, it should be said, the inner aspect of this is not unique to predictive processing.
36:21
SPEAKER_01: So it should be said that
36:23
SPEAKER_01: A lot of people writing on this topic have taken slightly different angles toward the same solution.
36:29
SPEAKER_01: So Shannon Valor, for example, has taken a kind of virtue ethics approach, a kind of Aristotelian.
36:36
SPEAKER_01: She talks about techno moral virtues and what kinds of character traits should we
36:41
SPEAKER_01: aim towards cultivating.
36:43
SPEAKER_01: And of course, all of this can be kind of integrated with the active inference story as well.
36:48
SPEAKER_01: And there's others as well that have said similar things.
36:52
SPEAKER_00: And man,
36:53
SPEAKER_00: just for you,
36:54
SPEAKER_00: just going forward,
36:57
SPEAKER_00: Ben,
36:57
SPEAKER_00: you said the third one is environmental scaffolding,
37:00
SPEAKER_00: which is so good,
37:01
SPEAKER_00: but I would say that was the fourth one,
37:02
SPEAKER_00: right?
37:03
SPEAKER_00: The third one is redeployment of precision.
37:06
SPEAKER_00: So you have update your model,
37:08
SPEAKER_00: act on the world,
37:09
SPEAKER_00: just change your precision patterns,
37:11
SPEAKER_00: which is attention control,
37:12
SPEAKER_00: right?
37:13
SPEAKER_00: Right.
37:13
SPEAKER_00: So, I mean, I love that.
37:14
SPEAKER_00: I mean, nobody's written that yet, that four way.
37:16
SPEAKER_00: I think we should, maybe we should do that.
37:18
SPEAKER_00: Cause that's like model updating on the world,
37:21
SPEAKER_00: changing the world and changing the way you deploy precision sort of is a,
37:25
SPEAKER_00: is a nice,
37:27
SPEAKER_00: nice four axes.
37:28
SPEAKER_01: Well,
37:28
SPEAKER_01: I think,
37:29
SPEAKER_01: I think niche construction might kind of,
37:33
SPEAKER_01: I'd be interested in how it compares formally to just like bog standard active inference,
37:37
SPEAKER_01: just kind of action perception loops.
37:39
SPEAKER_01: But you know, the,
37:41
SPEAKER_01: Yeah,
37:42
SPEAKER_01: I mean,
37:42
SPEAKER_01: taking into account the ways that you are fairly empowered in terms of how you
37:46
SPEAKER_01: structure your environment most of the time,
37:49
SPEAKER_01: even down to very simple things,
37:51
SPEAKER_01: you know,
37:51
SPEAKER_01: like leaving your phone in another room when you go to bed,
37:54
SPEAKER_01: setting reminders for certain things.
37:57
SPEAKER_01: You know, you can offload a fair amount of cognitive labor into the environment.
38:03
SPEAKER_01: Yeah.
38:05
SPEAKER_02: Yeah, definitely.
38:06
SPEAKER_02: And I think that kind of external...
38:10
SPEAKER_02: restructuring your environment.
38:11
SPEAKER_02: I've gotten very into the kind of deep hostile scaffolding idea that there's the,
38:16
SPEAKER_02: that they've,
38:17
SPEAKER_02: you know,
38:17
SPEAKER_02: that there's this like gambling technologies,
38:19
SPEAKER_02: I suppose there's certain online spaces that are set up.
38:22
SPEAKER_01: David Spurratt's work, I think.
38:24
SPEAKER_02: Yeah.
38:26
SPEAKER_02: Yeah.
38:26
SPEAKER_02: Is it?
38:27
SPEAKER_02: Yeah.
38:28
SPEAKER_02: Tim Spurratt.
38:28
SPEAKER_02: Tim and Spurratt.
38:30
SPEAKER_02: Not sure, but yeah.
38:32
SPEAKER_02: And that kind of,
38:32
SPEAKER_02: I mean,
38:33
SPEAKER_02: cause I accept,
38:34
SPEAKER_02: I think the meaning crisis framing for me,
38:36
SPEAKER_02: I talked to Verveke about it before as well,
38:38
SPEAKER_02: actually,
38:39
SPEAKER_02: in terms of this,
38:40
SPEAKER_02: Like you hear Mark Zuckerberg interviewed about Facebook and he said he wanted to connect people.
38:44
SPEAKER_02: And it was this idea that it was about connection.
38:46
SPEAKER_02: Nobody's connected anymore.
38:48
SPEAKER_02: So we're going to build these things to connect people.
38:49
SPEAKER_02: And there's a real narrative behind a lot of the structure of these things.
38:54
SPEAKER_02: Like they're very dramatic and they plug you into a story,
38:57
SPEAKER_02: I think,
38:58
SPEAKER_02: that makes sense of your life in terms of your status,
39:01
SPEAKER_02: your relationships to other people,
39:03
SPEAKER_02: who you are in the digital world,
39:05
SPEAKER_02: you know,
39:05
SPEAKER_02: what your brand is and
39:08
SPEAKER_02: I think it's filling a narrative void that we have in our lives in just this
39:13
SPEAKER_02: current meaning crisis society,
39:15
SPEAKER_02: that this is a substitute for people to understand themselves,
39:19
SPEAKER_02: to organize their attention and their experience.
39:22
SPEAKER_02: And so even just getting rid of social media, I feel like we'd invent something else.
39:26
SPEAKER_02: that would be just as bad like because addictions i think are those pseudo kind of
39:32
SPEAKER_02: they just they give you something to focus on but it's it's ultimately you know
39:36
SPEAKER_02: fruitless but i wonder do you do you see something then filling that void because i
39:40
SPEAKER_02: think that's what we're talking about in kind of this uh structuring of the
39:44
SPEAKER_02: environment like we need something else you know wisdom
39:49
SPEAKER_02: And institute wisdom, spiritual traditions.
39:51
SPEAKER_00: Yeah, we need to be frenemous.
39:53
SPEAKER_00: We need to be wise.
39:54
SPEAKER_00: I think we need to be wise people.
39:55
SPEAKER_00: And with Joan on that,
39:56
SPEAKER_00: I mean,
39:57
SPEAKER_00: the meaning crisis is vicious because,
39:59
SPEAKER_00: I mean,
40:01
SPEAKER_00: it's complicated,
40:01
SPEAKER_00: but the conditions that we find ourselves in have removed from us some of the
40:08
SPEAKER_00: ready-to-hand abilities that we used to have for dealing with existential terror.
40:13
SPEAKER_00: And now we are being regularly inundated with volatility that we don't know how to digest.
40:20
SPEAKER_00: And so then places that we can go where we get the sense that we're gripping,
40:25
SPEAKER_00: where we get a feeling like we're making it work,
40:28
SPEAKER_00: those become massively illuminated for us.
40:31
SPEAKER_00: I mean, it's really not a, it's not a surprise that some people are addicted.
40:34
SPEAKER_00: It's a, it's a surprise that we're not all addicted.
40:36
SPEAKER_00: Maybe we are,
40:37
SPEAKER_00: maybe that is,
40:38
SPEAKER_00: maybe that is something that we're experiencing now because there are huge,
40:41
SPEAKER_00: huge percentages of the population now that are hooked on social media in a,
40:45
SPEAKER_00: in a,
40:46
SPEAKER_00: you know,
40:46
SPEAKER_00: what we might typify as a pathological way,
40:48
SPEAKER_00: spending eight or nine hours a day is,
40:50
SPEAKER_00: is a lot.
40:51
SPEAKER_00: Um,
40:53
SPEAKER_00: They are alluring because we have lost some of the skills that we need to deal with the volatility.
41:00
SPEAKER_00: And at the same time, we have massive upticks in volatility.
41:04
SPEAKER_00: You know,
41:05
SPEAKER_00: Shannon Valor says you have this exponential increase in uncertainty with AI-based technologies.
41:12
SPEAKER_00: So we have this massive burst of uncertainty.
41:15
SPEAKER_00: And at the same time, we have a loss of our ability as humans to manage and digest and meet volatility.
41:23
SPEAKER_00: Put those two things together and we're adrift.
41:26
SPEAKER_00: Anything that gives you a sense today that you are not adrift,
41:30
SPEAKER_00: I mean,
41:30
SPEAKER_00: it becomes totally,
41:32
SPEAKER_00: I think it's completely a normal thing that we would reach out and find it there.
41:36
SPEAKER_00: So I think, man, you're exactly right.
41:39
SPEAKER_00: As long as those are the conditions,
41:41
SPEAKER_00: then we will keep finding,
41:43
SPEAKER_00: we will keep finding and creating and leveraging opportunities to feel better,
41:47
SPEAKER_00: even if we're not getting better.
41:49
SPEAKER_00: So I think,
41:50
SPEAKER_00: you know,
41:50
SPEAKER_00: the big,
41:51
SPEAKER_00: the big solution here is that we,
41:53
SPEAKER_00: we start maybe taking some bites out of,
41:55
SPEAKER_00: out of,
41:56
SPEAKER_00: um,
41:56
SPEAKER_00: out of what we think of as the basis for the mean crisis,
41:59
SPEAKER_00: which for me,
41:59
SPEAKER_00: and then I wasn't really expecting you to talk about this on the podcast,
42:02
SPEAKER_00: but I really think it is wisdom.
42:04
SPEAKER_00: Like, um, making, making, I think, um, Shannon Valor is right.
42:09
SPEAKER_00: I think if we were to turn towards thinking about, um, what do I want to be as a human?
42:15
SPEAKER_00: And if that's,
42:16
SPEAKER_00: if that's scaffold around virtue,
42:17
SPEAKER_00: um,
42:18
SPEAKER_00: and you,
42:18
SPEAKER_00: you set that at the,
42:19
SPEAKER_00: you set that at the center of your life,
42:22
SPEAKER_00: um,
42:22
SPEAKER_00: and you start developing some of the skills that that requires things like
42:24
SPEAKER_00: tolerance to uncertainty,
42:26
SPEAKER_00: being comfortable with being uncomfortable,
42:28
SPEAKER_00: um,
42:29
SPEAKER_00: being cool with failure,
42:30
SPEAKER_00: um,
42:31
SPEAKER_00: loving being awkward and socially weird, giving a shit about other people, serving your community.
42:38
SPEAKER_00: You put those at the heart, these things start falling away spontaneously because they are antithetical.
42:44
SPEAKER_00: A lot of these are antithetical to that primary work.
42:47
SPEAKER_00: So it's a,
42:48
SPEAKER_00: I think it might be a,
42:49
SPEAKER_00: ultimately it might be a goal,
42:51
SPEAKER_00: a goal issue that if you can,
42:53
SPEAKER_00: if we could change the way that we're thinking about the kinds of things that we
42:55
SPEAKER_00: want to be,
42:56
SPEAKER_00: maybe that starts already softening some of these edges,
42:59
SPEAKER_00: then we could maybe possibly use it.
43:01
SPEAKER_00: We could use it without being hooked by it.
43:04
SPEAKER_00: Um, because, um, because we're not so adrift that it's so alluring.
43:11
SPEAKER_01: Yeah.
43:11
SPEAKER_01: I think that's absolutely right.
43:13
SPEAKER_01: I think it's interesting to think about addiction,
43:17
SPEAKER_01: kind of drug and alcohol addiction as a kind of...
43:20
SPEAKER_01: I think a lot of addicts who get sober from substances,
43:25
SPEAKER_01: I think they experience a kind of localized meaning crisis,
43:31
SPEAKER_01: a kind of very acute meaning crisis,
43:33
SPEAKER_01: where it's not just that the drugs or the alcohol were kind of chemically nourishing,
43:38
SPEAKER_01: but actually...
43:40
SPEAKER_01: They're nourishing in the sense that they provide an identity and they provide a
43:44
SPEAKER_01: kind of playbook for like,
43:45
SPEAKER_01: oh,
43:46
SPEAKER_01: this is what I do in social situations.
43:47
SPEAKER_01: I just get I get really drunk and chaos ensues and I'm gregarious and I don't really care about anything.
43:54
SPEAKER_01: And so it's nourishing in a very, very deep way.
43:58
SPEAKER_01: It gives you a kind of expectation about who you are as a person and how you're going to act.
44:04
SPEAKER_01: And when you get sober and that's dragged away from you,
44:08
SPEAKER_01: part of the crisis there is not just,
44:10
SPEAKER_01: there's not just a chemical withdrawal,
44:12
SPEAKER_01: but there's kind of like a meaning withdrawal as well.
44:14
SPEAKER_00: Yeah, if I don't go to the bar, what do I do on a Friday night?
44:17
SPEAKER_00: I mean, what am I going to do?
44:18
SPEAKER_00: Play soccer?
44:19
SPEAKER_00: You can feel like that coming out of long-term addiction.
44:23
SPEAKER_00: Like, what am I going to do?
44:24
SPEAKER_00: Like, be a sport guy now?
44:26
SPEAKER_01: I mean... And part of the problem there is recognizing that the
44:31
SPEAKER_01: the path back from that is slow.
44:34
SPEAKER_01: And this is,
44:35
SPEAKER_01: it comes back to the supernormal stimulus stuff,
44:37
SPEAKER_01: which is,
44:38
SPEAKER_01: I think there's an interesting distinction here between fast and slow,
44:42
SPEAKER_01: which is,
44:43
SPEAKER_01: it's part of what needs to be cultivated.
44:45
SPEAKER_01: And I'm no expert on wisdom or these kinds of practices at all,
44:48
SPEAKER_01: but it seems to me that what we're trying to cultivate here is a kind of the
44:53
SPEAKER_01: competing tendencies that I was talking about.
44:56
SPEAKER_01: The ones that are worth cultivating are ones that have to be cultivated over time.
45:01
SPEAKER_01: and with care and with some ups and downs and, and stepping away from the expected.
45:06
SPEAKER_01: I mean,
45:06
SPEAKER_01: a lot of the work Mark's done on aerodynamics is about kind of tuning our
45:11
SPEAKER_01: expectations for prediction error minimization.
45:14
SPEAKER_01: And I think internalizing what's going on with aerodynamics is really is,
45:18
SPEAKER_01: can be really helpful because it's realizing that actually your expectations can
45:22
SPEAKER_01: shift over time.
45:23
SPEAKER_01: Okay.
45:24
SPEAKER_01: So the model that you have of the world that says like,
45:27
SPEAKER_01: if things aren't going fabulously well for me all the time,
45:30
SPEAKER_01: then I'm going to feel really bad and frustrated.
45:32
SPEAKER_01: That's a temporary state of being.
45:35
SPEAKER_01: That's not a necessary feature of human experience.
45:39
SPEAKER_01: And so I think part of the work is tuning those expectations to understand that the
45:44
SPEAKER_01: things that can be more nourishing,
45:47
SPEAKER_01: like real-world social connections and relationships and creative practices and
45:54
SPEAKER_01: things like that,
45:55
SPEAKER_01: they
45:56
SPEAKER_01: They simply don't offer these very steep gradients of prediction error minimization
46:01
SPEAKER_01: that some other things offer.
46:03
SPEAKER_01: And that's okay.
46:05
SPEAKER_00: Yeah.
46:06
SPEAKER_00: Yeah.
46:06
SPEAKER_00: Nice.
46:07
SPEAKER_02: And it might even be better for your long-term self.
46:10
SPEAKER_02: Well, for sure.
46:13
SPEAKER_00: Think about getting off sugar.
46:15
SPEAKER_00: Sugar's cuing to the body.
46:16
SPEAKER_00: Refined sugar's cuing to the body.
46:17
SPEAKER_00: Like, wow, you're really killing it.
46:19
SPEAKER_00: In a world where there are limited resources, that's what the biosystem thinks.
46:23
SPEAKER_00: Like, you're doing great.
46:24
SPEAKER_00: Like, McDonald's is, you know, you're hitting all the buttons.
46:30
SPEAKER_00: And to come off of that requires that you do a couple of things.
46:34
SPEAKER_00: First of all, you have to tolerate the uncertainty.
46:36
SPEAKER_00: that eating vegetables has less gradient to it.
46:40
SPEAKER_00: But if you take the time and readjust your palate,
46:44
SPEAKER_00: because we've ruined our palate,
46:46
SPEAKER_00: this is like we've ruined our predictive palate.
46:48
SPEAKER_00: If you give it enough time, the palate will adjust because we're flexible systems.
46:53
SPEAKER_00: And then you are back getting good gradients from actually healthy food.
46:58
SPEAKER_00: That I remember when I first became a vegetarian, I did it for health reasons.
47:04
SPEAKER_00: It took about three months maybe for my palate to sort of update.
47:07
SPEAKER_00: And then I loved it.
47:08
SPEAKER_00: And then I really loved it.
47:09
SPEAKER_00: And then the more subtle things that were available to me were delicious and wonderful,
47:13
SPEAKER_00: but it took the time to actually,
47:15
SPEAKER_00: you know,
47:16
SPEAKER_00: you're snow blind when you,
47:17
SPEAKER_00: when you make a transition like that,
47:19
SPEAKER_00: you have to hang around until the predictive system updates.
47:22
SPEAKER_02: Like destroying your fixed point attractors.
47:25
SPEAKER_02: And then you end up with an increase in uncertainty and then eventually you
47:28
SPEAKER_02: actually get a better model because of it.
47:30
SPEAKER_02: Like, I don't know that you talk about wellbeing and that that's the model really of like,
47:34
SPEAKER_02: the difference between like psychopathology and wellbeing is the ability to do that,
47:38
SPEAKER_02: go through that kind of process.
47:40
SPEAKER_02: And I think maybe for,
47:42
SPEAKER_02: yeah,
47:43
SPEAKER_02: just like,
47:44
SPEAKER_02: I suppose to go into some wild speculation in terms of,
47:47
SPEAKER_02: do you think there's tech solutions to the meaning crisis?
47:51
SPEAKER_02: Cause I know this kind of ties in with your ambient smart environments as well.
47:54
SPEAKER_02: Like,
47:54
SPEAKER_02: cause we're talking about metacognition,
47:56
SPEAKER_02: like using technology to afford us metacognition that you could use again to
48:01
SPEAKER_02: proportion your attention to
48:02
SPEAKER_02: better like if you had metrics of that like a lot of the metrics now are going the
48:07
SPEAKER_02: other way but it's like if you could measure you know like how well am i doing in
48:10
SPEAKER_02: my relationship how often am i practicing wisdom how much of my attention am i
48:14
SPEAKER_02: wasting on other stuff yeah for me it seems like there is an opportunity there for
48:19
SPEAKER_02: like anti-meaning crisis technologies nice yeah i like that we're optimists yeah
48:29
SPEAKER_01: yeah it's funny i mean because my this work grew out of it was like this ambient
48:33
SPEAKER_01: smart technology work is kind of reactionary against the social the social media
48:37
SPEAKER_01: stuff i felt like i was being pessimistic for so long and whenever i gave whenever
48:41
SPEAKER_01: i gave talks on that stuff people were always like oh you know like focusing too
48:45
SPEAKER_01: much on the negative and so i kind of went away and for the next project i thought
48:49
SPEAKER_01: you know i just want to like
48:51
SPEAKER_01: really kind of be imaginative,
48:53
SPEAKER_01: forget about the economic incentives,
48:56
SPEAKER_01: the business model stuff,
48:57
SPEAKER_01: forget about all of that.
48:58
SPEAKER_01: And just think what can we,
49:00
SPEAKER_01: given the image,
49:02
SPEAKER_01: the picture we have of the active inference agent,
49:04
SPEAKER_01: what can we do into what kind of technology can we imagine?
49:08
SPEAKER_01: And I think we can imagine some pretty amazing technology.
49:10
SPEAKER_01: And I think that there are no guarantees here.
49:14
SPEAKER_01: Obviously, there are material realities, social, political realities.
49:19
SPEAKER_01: But I think actually we can imagine really, really fabulous technology.
49:23
SPEAKER_01: And I think as we continue to do work on this,
49:26
SPEAKER_01: I think the aim,
49:27
SPEAKER_01: at least for me,
49:28
SPEAKER_01: is to begin to maybe carefully uncover,
49:31
SPEAKER_01: like an archaeologist,
49:32
SPEAKER_01: some principles so that we can develop maybe a principled approach to developing
49:36
SPEAKER_01: technology
49:37
SPEAKER_01: good technology.
49:40
SPEAKER_00: This might actually be a good place to just jump into that a little bit, Ben.
49:44
SPEAKER_00: Ben and I have a new paper out on Ambient Smart Tech.
49:48
SPEAKER_00: The first thing to say about it is it's different than the technology we've been
49:51
SPEAKER_00: talking about so far because you would own this
49:53
SPEAKER_00: This wouldn't be a vicious AI owned by a corporation bending you towards their profit margin.
50:01
SPEAKER_00: This would be an AI that you own.
50:03
SPEAKER_00: I don't know if people at home know what an ambient smart environment is, but this is Ben's forte.
50:08
SPEAKER_00: So I think that'd be kind of cool, Ben.
50:09
SPEAKER_00: Why don't we lay out a little bit of what we think?
50:12
SPEAKER_02: I really think it is.
50:14
SPEAKER_02: the opposite of the social media nightmare i'm glad you said that because that's
50:22
SPEAKER_01: that is kind of explicitly what it was supposed to be in terms of like the work i
50:26
SPEAKER_01: was doing it really was supposed to be just like a counterbalance for some of the
50:30
SPEAKER_01: previous work i'd done and um
50:33
SPEAKER_01: Yeah, so an ambient smart environment is an environment that you work backwards.
50:40
SPEAKER_01: So it's a classroom or a home or an office,
50:44
SPEAKER_01: which is infused with smart technology,
50:47
SPEAKER_01: which is kind of we all,
50:48
SPEAKER_01: you know,
50:49
SPEAKER_01: smart technology is kind of learning about us and updating itself kind of automatically.
50:54
SPEAKER_01: The ambient part is what I think is really interesting.
50:58
SPEAKER_01: I think that's where the key, the magic lies.
51:00
SPEAKER_01: Ambient technology is just technology that functions without a kind of direct human machine,
51:07
SPEAKER_01: without an intentional moment of interaction or interface.
51:10
SPEAKER_01: So a really easy example to get this is most of us who drive cars now use ambient
51:16
SPEAKER_01: technology where we have a key fob in our pocket.
51:18
SPEAKER_01: and it just unlocks the car when we walk over to the car.
51:22
SPEAKER_01: So you no longer have to press a button or put a key into the lock.
51:27
SPEAKER_01: Automatic doors, I think, are an example of ambient technology.
51:30
SPEAKER_01: They just have a sensor, and they kind of open the doors when you walk in.
51:36
SPEAKER_01: There are already kind of examples of ambient technology.
51:40
SPEAKER_01: It should be said, I didn't invent this kind of vision for technology.
51:45
SPEAKER_01: I think there was an essay all the way back in about 19...
51:50
SPEAKER_01: 91 by Mark Weiser on kind of Internet of Things invisible technology.
51:58
SPEAKER_01: I forget the precise phrase that Weiser uses,
52:01
SPEAKER_01: but essentially like computing power that's operating in the background and
52:05
SPEAKER_01: adapting in real time to a person's needs.
52:09
SPEAKER_01: And so that's when when we talk about ambient smart environments,
52:13
SPEAKER_01: that's basically what we're talking about.
52:15
SPEAKER_01: And most of the examples that we use in the papers are quite speculative examples.
52:20
SPEAKER_01: So I refer to this all the time as near future technology because it's.
52:24
SPEAKER_01: There's nothing here that we couldn't build in principle.
52:30
SPEAKER_01: It's just a question of taking the time for this to be adopted and picked up.
52:36
SPEAKER_01: There's already examples of medical research,
52:39
SPEAKER_01: for example,
52:41
SPEAKER_01: into care-assisted living homes for people with dementia that are making use of very...
52:47
SPEAKER_01: that highly network systems of ambient technology.
52:50
SPEAKER_01: Um, so I think it's just a matter of time.
52:52
SPEAKER_01: And I think that once you start thinking about it, you can get really, really imaginative with this.
52:57
SPEAKER_01: So,
52:57
SPEAKER_01: um,
52:57
SPEAKER_01: I don't know how long,
52:58
SPEAKER_01: how long you want me to carry on with this for,
53:00
SPEAKER_01: but,
53:00
SPEAKER_01: uh,
53:01
SPEAKER_01: like we,
53:01
SPEAKER_01: uh,
53:02
SPEAKER_01: we've,
53:02
SPEAKER_01: we've been thinking about the ways that like,
53:04
SPEAKER_01: uh,
53:06
SPEAKER_01: So something that's really... There are very obvious applications of this.
53:09
SPEAKER_01: So one example that already exists is a smart refrigerator, right?
53:14
SPEAKER_01: So you could have a refrigerator that...
53:17
SPEAKER_01: You kind of tell it that you want to get better at cooking and it starts monitoring
53:22
SPEAKER_01: the kinds of things that you cook.
53:24
SPEAKER_01: It's linked up to your oven.
53:25
SPEAKER_01: It's hooked up to your kind of your grocery store account.
53:29
SPEAKER_01: And it's kind of learning the kinds of things you tend to cook.
53:33
SPEAKER_01: It's maybe even using ambient sensors in the environment to watch how you cook.
53:38
SPEAKER_01: And then you tell it that you want to improve a steady rate.
53:41
SPEAKER_01: And so it starts to just kind of turn up the uncertainty and kind of bring in
53:45
SPEAKER_01: ingredients that it knows that you might be unfamiliar with.
53:49
SPEAKER_01: And I think the thing here that's really important to recognize,
53:52
SPEAKER_01: and this is a key feature of a key claim in the paper that we just published,
53:56
SPEAKER_01: is I think up until now with traditional technology,
54:00
SPEAKER_01: we tend to think of technology as simplifying a problem space.
54:05
SPEAKER_01: So technology is there to make life easier in some way for us.
54:08
SPEAKER_01: I think when we think about ambient smart environments,
54:12
SPEAKER_01: I think one thing that is very striking is that when we think in terms of attractor
54:18
SPEAKER_01: states and uncertainty gradients and how that connects to affectivity and
54:24
SPEAKER_01: well-being in the work that Mark's done,
54:26
SPEAKER_01: the biggest potential here is actually to have environments that can turn up the uncertainty.
54:31
SPEAKER_00: Yeah, they make it harder for you.
54:34
SPEAKER_01: environments technology that is not necessarily trying to make you reduce
54:38
SPEAKER_01: uncertainty immediately faster and so in the paper we talk about shifting us
54:43
SPEAKER_01: towards exploratory action policies shifting us towards epistemic affordances and
54:49
SPEAKER_01: epistemic actions technology that is driving you to learn and to and to learn new
54:56
SPEAKER_01: new embodied skills
54:58
SPEAKER_01: Um,
54:59
SPEAKER_01: and this has,
55:00
SPEAKER_01: um,
55:00
SPEAKER_01: there's another paper with Ines Cipollito where we talk about the,
55:03
SPEAKER_01: the,
55:03
SPEAKER_01: the,
55:04
SPEAKER_01: the potential for like genuinely therapeutic application.
55:08
SPEAKER_01: Um, because in, in, in, in, um, psychiatric research, there is work where
55:13
SPEAKER_01: where psychopathology is described through the language of complex systems,
55:18
SPEAKER_01: where,
55:18
SPEAKER_01: say,
55:19
SPEAKER_01: like an instance of psychopathology is a kind of attractor basin characterized by
55:23
SPEAKER_01: certain symptoms.
55:24
SPEAKER_01: And the goal of psychotherapy and psychiatry is to disrupt those attractors.
55:30
SPEAKER_01: And so what you want to do is kind of turn the heat up,
55:32
SPEAKER_01: turn up the uncertainty,
55:33
SPEAKER_01: disrupt habitual patterns of behavior.
55:38
SPEAKER_01: So that's like a key distinction.
55:41
SPEAKER_01: And then once you have that on the table,
55:42
SPEAKER_01: the refrigerator example,
55:44
SPEAKER_01: and there's lots of different examples you can think of that are fairly straightforward,
55:47
SPEAKER_01: but you can get really weird with this.
55:49
SPEAKER_01: So there's neuroscience research that shows,
55:52
SPEAKER_01: for example,
55:53
SPEAKER_01: the frequency of rhythm,
55:56
SPEAKER_01: like particular frequencies in lighting,
55:59
SPEAKER_01: And visual cues can actually entrain breathing patterns.
56:03
SPEAKER_01: So there's research that shows you can influence how quickly people walk through
56:08
SPEAKER_01: airports by modulating the frequency of particular visual cues in the environment.
56:13
SPEAKER_01: And so you can imagine now kind of furniture that detects it's hooked up to a Fitbit or something.
56:20
SPEAKER_01: It detects when you're stressed and anxious,
56:22
SPEAKER_01: and it just introduces some haptic rhythm into your chair or your sofa that
56:28
SPEAKER_01: entrains like a different breathing pattern.
56:31
SPEAKER_01: There's research that shows that improved interoceptive accuracy can reduce symptoms of psychopathology.
56:37
SPEAKER_01: So what about a room that can now essentially,
56:40
SPEAKER_01: in kind of more or less subtle ways,
56:42
SPEAKER_01: put you inside your own body?
56:44
SPEAKER_01: and kind of put you more in touch with your own heart rhythms and other
56:48
SPEAKER_01: interoceptive processes to increase to decrease the uncertainty between uh brain
56:53
SPEAKER_01: and body um so that's been i've been that's what i've been doing recently is a lot
56:58
SPEAKER_01: of my time is spent just kind of fantasizing about increasingly weird ways that you
57:04
SPEAKER_01: can utilize this feedback loop between the environment and and the agent and um
57:10
SPEAKER_01: To circle back, I think ambient technology has this huge potential.
57:17
SPEAKER_01: I've been speaking to somebody recently who has a company that builds technology
57:23
SPEAKER_01: for elder care settings,
57:24
SPEAKER_01: for care homes,
57:26
SPEAKER_01: people with dementia and people who care for people with dementia.
57:30
SPEAKER_01: A lot of the technology around dementia care is focused on like the cognitive problem.
57:34
SPEAKER_01: It's like,
57:35
SPEAKER_01: you know,
57:35
SPEAKER_01: the very obvious symptoms,
57:37
SPEAKER_01: people like with memory issues and problems communicating.
57:40
SPEAKER_01: But one thing anybody who has been confronted with dementia will know is that
57:45
SPEAKER_01: there's a serious affective side to this as well.
57:49
SPEAKER_01: There's a profoundly difficult emotional side.
57:52
SPEAKER_01: um side to dementia where people stay experience a lot of anxiety a lot of
57:57
SPEAKER_01: frustration and they have problems communicating that and it's very easy to think
58:00
SPEAKER_01: of the way that ambient technology and network devices could communicate could help
58:06
SPEAKER_01: alleviate those problems in various ways so i'm really excited about this this is
58:11
SPEAKER_01: this is pretty much all i'm working on right now and
58:13
SPEAKER_01: I think that the active inference framework and the connection between affectivity
58:17
SPEAKER_01: and uncertainty is really useful.
58:20
SPEAKER_01: And yeah,
58:21
SPEAKER_01: the ways that technologically infused environments could kind of tap into that and
58:26
SPEAKER_01: pick up some of like the predictive slack.
58:29
SPEAKER_01: and really help you.
58:31
SPEAKER_01: We talked about fastness and slowness before and trying to cultivate a particular
58:36
SPEAKER_01: topography of attractor states that is conducive to that flexibility that
58:41
SPEAKER_01: characterizes well-being.
58:44
SPEAKER_01: It's really easy to imagine the ways that ambient environments could really support
58:48
SPEAKER_01: agents in doing that to break up problematic habits or to introduce flexibility in
58:55
SPEAKER_01: strategic ways.
58:58
SPEAKER_01: so yeah this this is this is what i'm working on now and i so good man it's really
59:02
SPEAKER_02: exciting um i i really i i've read the paper but yeah no i'm looking forward to
59:06
SPEAKER_02: reading more in terms of that just seems to me like the positive vision for a lot
59:10
SPEAKER_02: of this stuff like that's the way it could go if people that actually you know gave
59:15
SPEAKER_02: a shit about other people were running it you know and i think that sometimes like
59:20
SPEAKER_02: yeah like the next stage because a lot of this stuff once it's developed it's
59:23
SPEAKER_02: cheaper it's easier to use like it does become more common parlance so um
59:28
SPEAKER_02: Yeah,
59:29
SPEAKER_02: I don't think it's an actually unrealistic vision to think that you could apply
59:32
SPEAKER_02: those things at some point once it's made.
59:34
SPEAKER_02: I don't know,
59:35
SPEAKER_02: Mark,
59:35
SPEAKER_02: sorry,
59:35
SPEAKER_02: I know you have to run now in a second,
59:37
SPEAKER_02: but yeah,
59:38
SPEAKER_02: is there anything you want to add to that about the ambient smart environments?
59:42
SPEAKER_02: No, it's awesome.
59:44
SPEAKER_00: I mean,
59:45
SPEAKER_00: the only thing,
59:46
SPEAKER_00: no,
59:46
SPEAKER_00: I'm not sure,
59:47
SPEAKER_00: but I'm just struck by how cool I think it would be if I had a AI assistant,
59:53
SPEAKER_00: personally owned AI assistant,
59:55
SPEAKER_00: who scheduled my meetings to end a little bit early on Tuesday and proposed an
1:00:00
SPEAKER_00: ingredient I've never used that needs to be cooked with a technology I haven't used yet,
1:00:04
SPEAKER_00: like a new kind of using a knife or use an air fryer in a way that I've never used before.
1:00:10
SPEAKER_00: And then it suggests that as a possible menu approach.
1:00:14
SPEAKER_00: So I'm given a little bit more time because the ambient smart tech knows I'm going
1:00:19
SPEAKER_00: to need more time to work this out.
1:00:21
SPEAKER_00: I just think that's so cool.
1:00:23
SPEAKER_00: I think it's such a cool thing to have a digital system.
1:00:25
SPEAKER_02: It's something I thought about with social media as well.
1:00:28
SPEAKER_02: I mean,
1:00:28
SPEAKER_02: if you could optimize your own algorithm,
1:00:30
SPEAKER_02: you'd probably want it to optimally challenge you rather than just keeping you in
1:00:34
SPEAKER_02: these feedback loops.
1:00:36
SPEAKER_00: Ideally, you'd want to have a balanced news source.
1:00:39
SPEAKER_00: I don't want anything where I'm just seeing inside a filter bubble.
1:00:41
SPEAKER_00: I want to be able to see from different perspectives that I,
1:00:44
SPEAKER_00: as the thinker,
1:00:45
SPEAKER_00: have the resources I need to think well.
1:00:49
SPEAKER_00: Yeah.
1:00:49
SPEAKER_01: Yeah,
1:00:50
SPEAKER_01: I think a really important thing to mention as well,
1:00:52
SPEAKER_01: I haven't already,
1:00:53
SPEAKER_01: is that talking about carefully uncovering principles that I think would apply to
1:00:59
SPEAKER_01: good technology,
1:01:00
SPEAKER_01: I think one that characterizes ambient smart technology is,
1:01:05
SPEAKER_01: in general,
1:01:06
SPEAKER_01: it doesn't show up as an affordance in the environment itself.
1:01:11
SPEAKER_01: So this is one thing we argue in the paper.
1:01:13
SPEAKER_01: It's kind of obvious in hindsight, but it's nice to have it kind of down explicitly, is
1:01:19
SPEAKER_01: The thing about social media and when we think about hyper stimulators and things
1:01:22
SPEAKER_01: is that what's really key there is the affordances that they offer to the agent.
1:01:28
SPEAKER_01: And in technology theory,
1:01:30
SPEAKER_01: generally,
1:01:30
SPEAKER_01: we tend to think of like,
1:01:31
SPEAKER_01: oh,
1:01:32
SPEAKER_01: we think in terms of affordances,
1:01:33
SPEAKER_01: like what does this technology afford to the agent?
1:01:36
SPEAKER_01: I think what's really striking about ambient smart environments is that they don't
1:01:40
SPEAKER_01: afford anything themselves.
1:01:43
SPEAKER_01: So it's the same technology as before.
1:01:46
SPEAKER_01: So it's refrigerators and cookers and
1:01:48
SPEAKER_01: exercise bikes and all of this, but it's what's happening there.
1:01:52
SPEAKER_01: It's happening on a higher level.
1:01:54
SPEAKER_01: It's happening upstream of the agent's field of affordances.
1:01:57
SPEAKER_01: It's like this kind of like cultivating,
1:02:01
SPEAKER_01: overseeing,
1:02:02
SPEAKER_01: orchestrating role that's being played,
1:02:04
SPEAKER_01: which I think still needs to be properly theorized.
1:02:06
SPEAKER_01: So I think that's,
1:02:07
SPEAKER_01: there's going to be some future work on what exactly is the best way to understand
1:02:11
SPEAKER_01: that role.
1:02:12
SPEAKER_01: But having a technology that is comfortable to fade away into the background, that's not trying to like,
1:02:18
SPEAKER_01: keep your attention on it at all costs and is playing like a fundamentally very different role.
1:02:24
SPEAKER_00: Yeah, rather nudging your attention to other cares and concerns rather than back to it.
1:02:29
SPEAKER_00: It's very interesting.
1:02:31
SPEAKER_01: Less selfish technology.
1:02:32
SPEAKER_00: Oh, that's good.
1:02:34
SPEAKER_00: I almost hear a company name in that, you know.
1:02:37
SPEAKER_00: Yeah.
1:02:38
SPEAKER_00: generous tech or something selfless selfless tech is that what you just said yeah
1:02:43
SPEAKER_00: yeah i said generous but you know that's better maybe even just take out the
1:02:47
SPEAKER_02: negative inclination cool yeah i'm very exciting i'd love to talk to you again
1:02:53
SPEAKER_02: about all this to be honest because um yeah these are such great topics and you're
1:02:57
SPEAKER_02: both so insightful um so thank you so much uh for making time my pleasure really
1:03:01
SPEAKER_02: appreciate