The Ad Hominem Inference

One of my favorite academic papers of all time is Glenn Loury’s 1994 essay on Self-Censorship in Public Discourse. There isn’t a single equation there but the work is mathematically deep, as evidenced by the kinds of technical successors that it has inspired. It’s also beautifully written and deeply interdisciplinary, building on Erving Goffman’s pioneering work on impression management, and using Julius Caesar for illustration.

Brutus, argues Glenn, makes a “naive, guileless, literal” speech defending the assassination of Caesar in Act III of Shakespeare’s play. This seems to meet with approval from the assembled crowd, until Antony responds with a “powerfully manipulative oration” that makes “the words honorable man in reference to Brutus mean exactly their opposite.” We know how that story ends.

Glenn and I discussed the paper at length on his podcast a decade ago, and returned to the theme earlier this year. It was the topic of my presentation at a conference in his honor in 2022. The piece will be republished as a short book soon, with a new foreword and afterword by the author, a simple title, and a striking cover.

A central concept in the essay is the ad hominem inference, which Glenn defines as follows:

Ad hominem inference, though denigrated by the high-minded, is a vitally important defensive tactic in the forum. When discussing matters of collective importance, knowing “where the speaker stands” helps us gauge the weight to give to an argument, opinion, or factual assertion offered in the debate. If we know a speaker shares our values, we more readily accept observations from him contrary to our initial sense of things. We are less eager to dismiss his rebuttal of our arguments, and more willing to believe facts reported by him with unpleasant implications. The reason for all of this is that when we believe the speaker has goals similar to our own we are confident that any effort on his part to manipulate us is undertaken to advance ends similar to those we would pursue ourselves. Conversely, speakers with values very different from ours are probably seeking ends at odds with those that we would choose, if we had the same information. The possibility of adverse manipulation makes such people dangerous when allowed to remain among us undetected. Thus, whenever political discourse takes place under conditions of uncertainty about the values of participants, a certain vetting process occurs, in which we cautiously try to learn more about the larger commitments of those advocating a particular course of action.

Note that Glenn refers to inferences rather than attacks. He considers such reasoning to be motivated by self-protection, and perfectly consistent with human rationality.

But when judgements about values and character are made based on the content of speech, dissent from a widely shared consensus can become very costly, resulting in “social ostracism, verbal abuse, extreme disapproval, damage to reputation, and loss of professional opportunity.” These costs are most severe for those who do, in fact, share the values and commitments of the community; they may not matter at all for others. As a result, certain public speech acts are avoided by people who would like to remain in good standing, while being adopted with relish by those unconcerned with community approval.

The result is self-censorship and a hardening of orthodoxies:

For every act of aberrant speech seen to be punished… there are countless other critical arguments, dissents from received truth, unpleasant factual reports, or nonconformist deviations of thought that go unexpressed, or whose expression is distorted, because potential speakers rightly fear the consequences of a candid exposition of their views. As a result, the public discussion of vital issues can become dangerously impoverished.

As Glenn puts it in the foreword to the forthcoming book: the problem of censorship is far more subtle than commonly assumed, “entailing as it does not only the iron fist of state repression but also the velvet glove of social cooptation.”

Let me illustrate with a topical example.

In July of last year, while running for the Democratic nomination, Robert F. Kennedy Jr made the following remarks in a conversation that was surreptitiously recorded:

We need to talk about bioweapons… I know a lot now about bioweapons because I've been doing a book on it for the past two and a half years… We've put hundreds of millions of dollars into ethnically targeted microbes. The Chinese have done the same thing. In fact, Covid-19, there's an argument that it is ethnically targeted. Covid-19 attacks certain races disproportionately. The races that are most immune to Covid-19… because of genetic differentials… of the ACE2 receptor… Covid-19 is targeted to attack Caucasians and Black people; the people who are most immune are Ashkenazi Jews and Chinese… We don’t know if it was deliberately targeted… but there are papers out there that show the racial and ethnic differential impact… We do know that the Chinese are spending hundreds of millions of dollars developing ethnic bioweapons and we are developing ethnic bioweapons… that's what all those labs in the Ukraine are about, they’re collecting Russian DNA, they're collecting Chinese DNA, so we can target people by race.

There is a lot in this bizarre set of claims that one could take issue with, but it’s important to first understand that there is indeed a Cleveland Clinic study that examined genetic differences across populations in susceptibility to the disease. As noted by Paul Offit in his critique of Kennedy’s remarks, this study predicted during the very early days of the pandemic that “the groups most susceptible to Covid-19 were Africans, African Americans, and non-Finnish Europeans; those somewhat less susceptible were Latino, East Asian, Finnish, and South Asians; and those least susceptible were the Amish and Ashkenazi Jews.”

However, as explained very clearly by Offit, this study is not relevant to our understanding of population-level differences in fatality rates:

Data were collected and analyzed well before SARS-CoV-2 started killing people in the United States. For that reason, researchers didn’t correlate genetic susceptibilities with clinical outcomes. They were just predicting who they thought would be most likely to suffer from Covid-19… the maximum frequencies of genetic variations among different groups was no greater than 1 in 100… while it was reasonable to predict that one individual might be more susceptible to Covid-19 than another, these genetic differences were far too rare to account for population differences. Now that the virus has been circulating for almost four years—and killed about 7 million people worldwide—we know that their predictions were wrong. Hospitalizations and deaths weren’t determined by racial or ethnic backgrounds, they were determined mostly by age, underlying health problems, and vaccination status.

So Kennedy misunderstood or mischaracterized a study in order to support a narrative about heinous military activities by major powers, including our own.

But most headlines and social media commentary at the time focused not on factual error or baseless speculation, but on bigotry, with particular attention to anti-Semitism. There was some vigorous pushback but this was largely drowned out in the cacophony.

This is the ad hominem inference at work. It is a perfectly understandable impulse, for reasons explained in Glenn’s essay. But it has some unintended political consequences.

Once a public determination has been made that someone does not share the values and commitments of a community, a bridge is burned and the prospect of cooperation to meet shared goals becomes all but impossible. Even a simple act of courtesy such as taking a phone call becomes difficult to countenance. And this kind of disrespect can drive the apostate into the arms of a different, more welcoming community.

A month before Kennedy’s remarks came to light, I argued that “if his party adopts a dismissive and contemptuous stance towards him and towards those whom he has mobilized, it will sink its own prospects.” I stand by that assessment. Self-censorship and the emptiness of public discourse is not the only consequence of ad hominem reasoning. At certain critical historical junctures, the impulse can alter the path taken by a nation.

Three forthcoming titles from Polity Books.