Why Critical Thinking Is Not The Answer To Misinformation - Rob Peace

cheers um yeah so I I'm Rob and let's talk about why critical thinking is not the problem uh it's not the answer to tackling this and misinformation so I'm a social psychologist so I I care much about how the environment and people around us affect our Behavior so I'm not the other type of psychologists or listen to problems because I'm I'll give you rubbish advice and I don't really care um it's the truth um so I'm actually part of the center for doctoral training in trust identity privacy and security at scale which is collaboration between universities are back in Bristol and we're very infrastructure based on our security so my infrastructure is a little weird so

I'm trying to help secure um information communication online and our ability to trust that information so kind of like somebody of a talk so I'm going to present two of my first PhD um studies which are quite exploratory but they're quite fun hopefully so what I care about is why do people end up not that believing like um that covet could make you magnetic or get in this far where you becoming extremism and get arrested for the rest of your life and so what the process is why do normal rational people end up thinking like this uh so first thing to say is that we can't know everything and so we have to rely on other people

we have to trust other people's information and as you all know there's a lot of information online some trustworthy some not trustworthy so what I kind of care about is if you don't know what the information if you don't know if it's trusted or not how do we evaluate it and I say it is what not just me research says there's trust cues so we take these little heuristics from the environment the digital environment and we calculate the trustworthiness of information through those so what uh trust use and there's loads of different types there's anything that we use as a shortcut to trust information so it's been mentioned already social proof like rating systems and comments are quite a good one and

there's symbols like this so a living wage employer so that could show that the company is has a lot of Integrity so you might trust it because of that or there's ability to use and so adopters are often trusted um and even symbols like padlocks even if they should or shouldn't be they're still trusted so we take these um cues and people who want to be trusted try and communicate these cues across uh social media which is a noisy massive environment so the theory I kind of put to it is something called signal protection Theory which is really basic really simple and if the trust queues communicate properly it hits so this is a signal yes has the

response been good yes so I is someone who's reading the information and trusts it I've gone through that process other things can happen when it's noisy so there's false alarms so maybe you trust something you shouldn't or because it's a noisy environment you don't always get the messages you should so this is the kind of basic Simple Theory we were putting towards the um review which I'm about to talk about foreign

I did a systematic review which was taken a ton of papers so in the end it was two and a half well almost two and a half thousand Pages um going through a screening process and having 63 left that met all my criteria for the studies and from those 63 papers I looked at the evidence base for trust you so cues that we know will communicate and trustworthiness so even if it's a noisy environment they still works um so I got these studies from social technical disciplines so psychology um computer science uh electrical engineering a massive range um and really tried to hit the kind of social technical um space so we could synthesize all of

the evidence we have I also did a second question where I looked at interventions to help us evaluate these cues because as somebody ever talks and said we're not actually very good at evaluating what we should and shouldn't Trust online uh so it ended up looking like this so the results I did something called a thematic analysis which is a very qualitative method so I took all these cues I looked at them I coded them for what they actually did and what they were and then I grouped them together so the three themes I got from those groups uh one was verifiability to use so it was any information that believed people could verify who the person was and what they

said so profile pictures were a really good one but business names um email addresses however as someone said previously all stuff it could be really easily faked online so it's actually a bit useless uh expectancy violations so this was quite interesting is it's more about the digital environment and the design so as long I social media looked how people thought it should it was only when something kind of looked at a place that they started going through a more uh distrusting and critical not critical thinking so I'm having a go at that but maybe a more Suspicious Mind so as long as it looks right you're good and then the last one quickly talked about social aspects of trust so again

it was very social proofy very um comments likes retweets all these kind of things all these little heuristics that we use but are actually again pretty bad second one I looked at interventions that helped correctly judge digital information so we wanted to do a matter analysis and look at the effect size so we look at how the interventions were scored before they took part and then how they solved after however because it was this big range of disciplines report and standards are different across disciplines we weren't actually able to do that so again we have to go down the Thematic group so I'll just highlight a few so the technical ones often try to summarize

metadata and give you more information to evaluate even though you're starting up the skills so get in one way if you understand what the metadata is and other things but not so good if you don't social um interventions try to teach critical thinking and these the effect sizes and the uh how long the interventions last or tend to drop off after a few weeks so they're not actually that good anyway the one that was actually a really good one was this social technical intervention so what they did is put a midwife in a Facebook forum and had new mothers um have a direct link with expertise essentially and trust with the information so it was using expertise

which is something we should trust and using the system to distribute that knowledge properly so these are the kind of things where we should be taking advantage of um so did this big review of trust um trust queues had a big database now saying right I know these things raise trust but I know it's not simple as that um so I started trying to think okay how are these trust queues used so we all see the world differently which is evidence by this bit of Rugby's here so that is whales should be scoring a try a few years ago and my engagement to this day still says it wasn't a try and I'm like what clearly was but it's just trying to

demonstrate that your background could shape the perception of the world um and it definitely happens to behaviors so this model on the right is called the Brunswick Lens model so it's quite an old Theory but it describes um how when we don't know what the validity of information would be we take cues from across the environment so if it was on Twitter for example you could look at shared followers whether they have an avatar as a profile picture um previous posts you could look through we take these cues and these cues become a lens to how we uh validate the information and how we see the information whether it's trustworthy or not taking that kind of thinking we were

like okay well how do we explore this what do we do like how do we know if people are actually going through these processes so this is where study 2 comes in and this was um identifying trust using an open source software Library so we took the um our statistical Chrome language training when it doesn't know we looked at open source software libraries because I am technically a security researcher even though I'm very psychological based so we were like okay let's look at open source software attacks and let's start thinking about how people trust which packages what to download what code to address so for this we um had 20 participants they all had to

have used R and cram as a library before um however 10 were really experienced 10 were not so experienced so the methodology we went through we had a Microsoft teams meeting they shared their screen while it's interacting with Cran and basically they were pointing out to me things they trust things they didn't trust we then transcribed those interviews and coded it line by line to say this means this this means this this means this we don't have an even bigger um database of trust queues which we went on to again theme and categorize very similar before however slightly different applications so on the top left factors of trustworthiness which is classic trust Theory um which says ability benevolence and

integrity will raise if you have those friends or Raise Your perception of trust so these cues are things people were looking for without knowing they were looking for it um which for example was research papers anything that Quran any documentation that said actually this isn't just a code this has been published blah blah they saw as a ability or you could assembly tag review as well again digital environment matters so if anyone's used Quran here you'll know it's the horrible website it looks like it's from the 90s which was good because it started making people question things and got them talking quite a lot but again if you want to raise people's trust just make it look professional make it

look good and they just won't really question it this one's more important it's social aspects of trust um which is looking at um how we get recommendations from colleagues and friends and peers problem is trust in those people they don't necessarily know what they're on about either so we end up over relying on trust from people who aren't validating the information and it all goes into this Theory called routine activity Theory which is criminological theory that talks about crime will happen when we have a motivated offender which is pretty much any OSS attack because the victims come to you you have to do no work they just download it um a lack of K-pop Guardian so someone

else has talked about this morning we haven't really got the policing resources online and the people we are going to for help don't know what they're on about either so all creates this um suitable victim as well and it might explain why there's such a reason uh rise in OSS attacks so what's it means critical thinking um it means we're already critically thinking it's not that people aren't thinking about what they're interacting with online it's that over here the interaction of psychology information systems and engagement business models be quite hard for us to properly understand what information we should be interesting also in the past we had people like teachers Librarians and news editors that would pick which information we saw

before social media now we don't have that we've all got access to the internet we've all got access to this raw information that we can't necessarily or we haven't got the skills to judge yet and then finally like I said the lack of guardians or get information Gatekeepers is something we I'd say is a big one so can we reintroduce these uh which goes on to my last slide which is just a kind of future thoughts and what I'm trying to do in my research so we need to understand more about Trust online we um especially trust in information online a lot of our theories based on face-to-face trust which does change online from my research

um I need to do more quantification so we need more more statistics better measurement tools um to do better experiments then reintroducing the role of information Gatekeepers and online settings so this is more of a kind of business side but can we use people like midwives can we make the case that we should have expertise at your fingertip as well as having to go into you know physical places um and maybe it would save people money so I don't know Midwife instead of having to go to 10 different places in the town in the day they come to you online so there should be a case for it and then can we design an additional

environment to promote distrust so like I said earlier that horrible sight from the 90s did set people off it did make them think oh I don't trust this so what can we do to try and promote cues like that right and all the same Grandma who thinks this is a bird dog but it's a dog with a ham on his face or to kind of sort that out and that's kind of my key pitch yeah cheers foreign