Here’s what we’re all desperately trying to explain: certain groups seem to have a wildly over-inflated devotion to their beliefs. Such groups seem to flourish online. Their members seem to resist very good contrary evidence, and to slowly become more and more immune to criticism … [please read below the rest of the article].
Nguyen, C. Thi. 2019. “Group-Strapping, Bubble, or Echo Chamber?” Social Epistemology Review and Reply Collective 8 (6): 31-37. https://wp.me/p1Bfg0-4dr.
The PDF of the article gives specific page numbers.
- Boyd, Kenneth. 2019. “Epistemically Pernicious Groups and the Groupstrapping Problem.” Social Epistemology 33 (1): 61-73.
- Baumgaertner, Bert. 2019. “Groupstrapping, Bootstrapping, and Oops-strapping: A Reply to Boyd.” Social Epistemology Review and Reply Collective 8 (6): 4-7. https://wp.me/p1Bfg0-4bE.
Here’s what we’re all desperately trying to explain: certain groups seem to have a wildly over-inflated devotion to their beliefs. Such groups seem to flourish online. Their members seem to resist very good contrary evidence, and to slowly become more and more immune to criticism. And they group members seem to feed off of each other. These groups go epistemically toxic with stratospheric intensity, and with a force that individuals rarely seem to manage on their own. And that toxicity seems to be infectious. Normal, apparently rational, people get captured by such groups. Reasonable-seeming people walk in, and then they come out dogmatic and immovable. So: what in the hell is going on? Some explanations are in terms of individual psychology, like our individual tendency towards belief polarization in any form of reasoning. But other explanations focus on the essentially social, communal aspect of forming belief. Those essentially social explanations are what we’re considering here.
Epistemically Pernicious Groups
There are a few standing social explanations for the ferocity of these, as Kenneth Boyd calls them, epistemically pernicious groups. One explanation is in terms of an epistemic bubble. An epistemic bubble is a social structure in which the members aren’t exposed to all the right information. This should be an extremely familiar, and somewhat notorious, picture by now. In his influential book, The Filter Bubble, Eli Pariser drew us a picture of how easy it is to get trapped in such a structure, especially in today’s technological environment. You choose your friends on social medial; they all share your politics. So you are always exposed to arguments for your side, and rarely hear the arguments and evidence on the other side. What’s more, the algorithms which mediate our online experience invisibly personalize that experience. Google Search knows who you are, has an intricate profile of you based on all your past searches and clicks, and tailors its results to give you what you want. Which means, without your knowing it, your online experience has been filtered. But this isn’t a new phenomenon. As Bill Bishop pointed out, social mobility has been letting people self-sort into like-minded communities, long before the Internet came around. And geographic self-sorting can reduce exposure to the other side, too.
Another explanation is in terms of echo chambers. An echo chamber is a social structure that manipulates who its members trust. An echo chamber member may have plenty of exposure to people from the other side, but that echo chamber member has been brought to systematically distrust all outsiders. The echo chamber isolates its members, not by cutting off the lines of communication, but by manipulating credentials. And once members’ trust has been properly subverted—once they only trust other echo chamber insiders—then there’s no limit to how much confidence they might get. Outside of echo chambers, we are constantly running into disagreements and surprising evidence from sources we trust. All that disagreement enforces a certain degree of humility; we can only go on so far before we are forced to confront an error we’ve made, to revise our beliefs. But once an echo chamber has made its members distrust all outsiders, and only permits people on the inside who buy into that echo chamber’s key belief systems, then the insider’s confidence levels can skyrocket. There’s nothing to hold them back.
If this sounds to you lot like how cults work, then you’ve got exactly the right idea. In Jamieson and Cappella’s landmark study of the right wing echo chamber around Rush Limbaugh and Fox News, they point out that the echo chamber works my systematically alienating its members from the outside world.
To my mind, echo chambers are the real culprit of what’s going on. That whole “post-truth” phenomenon that’s going on? Climate change denial, anti-vaccination, and all that jazz? It’s not that people are stupid, or brutally irrational. And it’s not that people simply aren’t exposed to the other side. As plenty of recent studies have demonstrated, political extremists of both sides have plenty of access to, and are aware of, the arguments and evidence accepted by the other side. It’s that climate change denialists have systematically been brought to distrust science—they have accepted something like a conspiracy theory, where the whole edifice of climate change science is in the pay of corrupt liberal elites.
In a recent paper, Kenneth Boyd has identified what looks, to my eyes, like a distinctive and novel factor, to help us explain these phenomena, which he calls “group-strapping.” Group-strapping, says Boyd, “is a general phenomenon that exists between a member of a group and a group itself, in which a member illicitly increases her warrant in a belief by appealing to the testimony of a group, one whose epistemic position towards a proposition she is at least partially responsible for determining” (8-9). In Boyd’s picture, what happens is that your testimony helps to constitute the group’s testimony. This opens up a pernicious possibility. For you may be tempted to use that group’s testimony to bolster your own belief. But that group’s testimony is, in part, partially constituted by your own belief. So if you used that group’s testimony to bolster your own belief, you would be involved in a problematic loop—you are laundering your belief through the group, to produce an apparently independent form of testimony, which you use to bolster your own belief. But what you’re using to bolster your belief is, in part, made up of your own belief.
Notice that Boyd’s explanation has nothing to do with lack of exposure to other people, and nothing to do with manipulated trust. It has to do with a very particular kind of mistake: a metaphysical mistake, where you don’t realize how much your voice contributed to the group’s voice. It is, then, an entirely distinctive form of explanation for the political hell-scape.
As Boyd points out, group-strapping is a very special version of the problem of, as Tom Kelley puts it, double-counting the evidence. Imagine, as Wittgenstein suggests, somebody confronting a stack of identical newspapers, each with the headline, “Polar bears are now officially an endangered species.” Imagine that somebody flips through this stack of newspapers, and treats each individual copy as a distinctive bit of evidence. Imagine that they then raise their confidence level a little bit more, for each identical newspaper they uncover. This is obviously a mistake. Here’s the problem: We should, indeed, raise our confidence level for each independent piece of evidence we confront. If all five of us separately did a particular bit of arithmetic, and we all arrived at the same answer, then our collective agreement means something. Each additional agreeing voice should, in this case, increase our confidence level. But if all five of us believe that Elon Musk has gone bankrupt because we all read the very same Tweet this morning, then our agreement adds nothing extra. We are, in this case, like that stack of identical newspapers: our beliefs all come from a single source, and so each additional piece of testimony presents no extra evidence.
This introduces a particular possibility of error: of erroneously thinking that our testimony is independent, when it actually isn’t.
Similarly, imagine that we are working on our calculus problem sets. After I finish my work on the first problem, I glance over and see that you’ve come to the same conclusion. But, unbeknownst to me, you actually copied my work while I was on my phone. (I was tweeting angry hot-takes about Musk). So what I thought was a useful independent confirmation was, in fact, just an echoing of my own work. If I increased my confidence because I thought your work was an independent confirmation, I would be making a mistake: again, double-counting the evidence. What I think is your independent testimony is, as a matter of fact, a repetition in disguise.
Double-Counting Our Own Testimony
Boyd’s very interesting contribution is to identify another form of double-counting—one in which we double-count our own testimony as it appears inside group testimony. It goes like this: let’s suppose that I am a member of the local Anarchist Collective. I decide, for my own reasons, that Elizabeth Warren won’t get my vote, because she’s doesn’t have a sufficient history of promoting anarchist values. I go to the group meeting, where I see other four members of the Anarchist Collective. We take a vote, and we agree to adopt the resolution that Elizabeth Warren is a bad anarchist.
Suppose that I now turned around and took the fact that our Anarchist Collective adopted that resolution as additional evidence that Warren is a bad anarchist. The worry is that here, I’ll be double-counting. I’m in danger of counting my own evidence twice: once in the original form, and then illicitly, again, smuggled back to me under the header of a “group resolution.” In other words, the worry is that I will problematically treat the Anarchist Collective’s groupvoice as an independent source of evidence from my own, forgetting that my own voice is actually part and parcel of that group voice.
And this opens the door to a particularly a vicious upward spiral. If I increase my own individual degree of belief because of the Anarchist Collective’s resolution’s, and then that leads us strengthen the Anarchist’s Collective’s degree of belief in their resolution, and then I use that to increase my individual degree of belief, I am entering into a problematic and potentially endless upwards loop.
Is this really a problem? First, let’s take a step back. I don’t think it’s always bad for me to give a belief extra weight based on the testimony of a group of which I am a part. Groups can make judgments, have beliefs, and form intentions, and these group attitudes can contain an epistemic force that goes their individual constituents. Let’s follow, as Boyd does, in the footsteps of Margaret Gilbert, Phillip Petit, and Christian List here, and distinguish betweena merely aggregative group attitude and an integrative group attitude. A group has an aggregative attitude if all the individual members happen to have that attitude. But an aggregative attitude doesn’t capture what we want to capture when we attribute attitudes to groups. Suppose, for example, that every member of the American Medical Association happens to believe that climate change is real, and happens to love Vietnamese food. It would be wrong to attribute to the AMA as a whole a belief in climate change, or a love for Vietnamese food. Something more needs to happen before such attitudes can ascend of an attitude that is genuinely attributable to the group.
We do attribute a belief that vaccines work to the AMA. Why? It isn’t just that all the members of the AMA hold that belief. That belief has also worked its way through a very particular, and official set of group procedures. It’s gone through the sub-committees, been ratified through the right processes, and gone through the official votes. We attribute the pro-vaccination belief to the AMA, but not the climate change belief to the AMA, because only the former has gone through the designated procedures that make that belief the group’s. The existence of these processes and procedures are what constitute a group’s special attitudes, as something above and beyond just what its members mostly happen to hold.
Weighing Group Testimony
So is it always wrong to weight the testimony of a group, of which I am a part? Not necessarily. In cases of integrative group testimony, something counts as group testimony if it passes some procedure. In some cases, the very fact that it’s passed that procedure does, in fact, give that testimony some extra epistemic force—when the procedures themselves add some epistemic substance. Suppose, for example, we are part of a laboratory, in which we are supposed to evaluate the evidence according to a very specific procedure. Suppose that procedure is, in fact, is a method of double-checking our results, and passing the procedure does add some real assurance that our results are good. In that case, there is an epistemic difference between all of the individuals believing something, and the group’s believing it.
Similarly, imagine that our Anarchist Collective has a very specific and useful procedure for group discussion and deliberation, where we make sure to voice each of our concerns, and allow each person to respond to the concerns, and where we keep track of all the concerns on a special chalkboard, to make sure that we’ve dealt properly with everybody’s concerns. Going through that procedure adds something epistemically; it is a good form of deliberation, and the results of it give us something above and beyond our beliefs, going in. So: integrative group beliefs are those that have passed some special procedure. Insofar as that procedure is epistemically productive, then the fact that the group believes something doesadd extra epistemic weight to the mix.
So if I treat the group’s testimony as adding extra weight, and I keep that weight in line with what the procedure adds, then I’ve done nothing illegitimate. The integrative process of becoming a group belief does, in some cases, add new evidence—the evidence of having survived some group process of deliberation.
Group-strapping, then, turns out to be a very clear error when it exceeds that added weight. Thus, in order to problematically group-strap, I have to do the following:
1. I need to be part of an integrative group.
2. I need to believe that p.
3. My beliefs need to be partially constitutive of that group’s belief that p.
4. I need to use the fact that the group believes that p to increase my degree of confidence in p by some amount, x.
5. x needs to exceed the contribution made by the integrative procedure of the group.
Problematic group-strapping happens, not in any case when a member weights their group’s belief. It happens when that weighting outstrips the actual contribution of the integrative procedure, whereby individual beliefs are integrated into a group belief.
So: I think Boyd’s description of problematic group-strapping is perfectly coherent, and, if it occurred, it would be a genuine epistemic problem. But does it actually happen? Remember, there are a number of very different mechanisms by which groups could turn epistemically pernicious. Let’s review:
- Epistemic bubble effects: members of the group aren’t exposed to relevant evidence.
- Echo chamber effects: members of the group distrust outside sources.
- Group-strapping effects: members of the group have double-counted the testimony of the group.
Notice that it’s important to get right what’s going on, because these are descriptions of very different problems. Each requires a different intervention. If it’s an epistemic bubble that’s the problem, we need to expose people to the information they’re missing. If it’s an echo chamber effect, we need to restore trust. If it’s a group-strapping effect, we need to point out to people that they’re making a complicated metaphysical mistake, and not recognizing that their own beliefs are the partial constituents of the group of which they are apart. Of course, they could all be happening at once. But it seems very important to figure out which are the most significant players. I think we have some empirical evidence that strict epistemic bubble effects are less commonthan we might have thought. I also think we have empirical evidence that echo chamber effects have played a significant rolein the political landscape. What about group-strapping effects? What would count as evidence?
Political Landscape and Group-Strapping
For my own part, while I am certainly convinced that group-strapping is a theoretical possibility, I’m not yet convinced that group-strapping is one of the major players on the political landscape. Here’s why. First, let’s imagine how group-strapping works in very large groups. Suppose I am part of a ten-thousand person group. That group believes that, say, Elon Musk is a genius. Before I join that group, the testimony has a certain impact on my degree of belief. This is, as yet, not a form of group-strapping, because it’s entirely independent of me. (Of course, it would be a problem to individually count the testimony of all ten-thousand members of that group, and then count again the group’s testimony with the force of a ten-thousand person group, but that’s a simple and familiar case of double-counting).
Now I join the group. I add my voice to the choir. I might, at this point, group-strap. And that would, in a technical sense, be a mistake. But in order for that to be any significant impact on my degree of confidence, my added voice would need to move the needle significantly—it would need to change the force or weight of the group’s testimony enough to significantly change the impact the group’s testimony had on me. But it seems, on first glance, that adding my voice to the choir should have an almost negligible effect on the needle. I am just one among ten-thousand. Because remember, the amount of support I’m importing through the group, from the other ten-thousand people, is perfectly legitimate. The group-strapping problem comes from the amount that of epistemic my voice adds, which then gets laundered back to me through group testimony.
Of course, my added voice would move the needle significantly if we were talking about very small groups. Certainly, when we’re talking about two or three person groups, my voice matters. But in that case, it seems less and less plausible that individuals would actually make the group-strapping mistake. In small groups, my contribution to the group seems utterly transparent.
Of course, this is all just speculation. Perhaps people do significantly up-weight group testimony when they have added their own voice, and perhaps people do make the group-strapping mistake even in tiny groups. What we need, at this point, is some empirical evidence. We philosophers have mapped the conceptual space—now we need to convince some social scientists to use that conceptual map to examine the real world.
Boyd, Kenneth. 2019. “Epistemically Pernicious Groups and the Groupstrapping Problem.” Social Epistemology 33 (1): 61-73.
Nguyen, C. Thi. 2018. “Echo Chambers and Epistemic Bubbles.” Episteme. https://doi.org/10.1017/epi.2018.32.
Nguyen, C. Thi. 2018. “Escape the Echo Chamber.” Aeon, April 9. https://aeon.co/essays/why-its-as-hard-to-escape-anecho-chamber-as-it-is-to-flee-a-cult.
Jamieson, Kathleen Hall and Joseph N Cappella. 2008.Echo Chamber: Rush Limbaugh and the Conservative Media Establishment.Oxford: Oxford University Press.
Kelly, Thomas. 2010. “Peer Disagreement and Higher Order Evidence.” In Social Epistemology: Essential Readings, edited by A. Goldman and D. Whitcomb, 183–217, Oxford: Oxford University Press.
Kelly, Thomas. 2008. “Disagreement, Dogmatism, and Belief Polarization.”The Journal of Philosophy 105 (10): 611-633.
List, Christian and Philip Pettit. 2011.Group Agency: The Possibility, Design, and Status of Corporate Agents. Oxford: Oxford University Press.
Pariser, Eli. 2011. The Filter Bubble. New York: Penguin Books.