by Nate Sheff
We trained our dog Gemini using positive reinforcement techniques, “clicker training.” She can sit, give handshakes and high fives, roll over, and we never used negative reinforcement to teach her.
That’s all true, but what do I mean by negative reinforcement? A lot of us assume that it has something to do with punishment, whereas positive reinforcement involves rewarding good behavior, but this assumption isn’t strictly true. To reinforce a behavior is to make that behavior more likely. Giving Gemini a treat when she sits after I make a certain gesture positively reinforces the association between my gesture and her sitting, because it makes that behavior more likely next time I make the gesture. I’m introducing something Gemini wants to strengthen a particular connection between my gesture and her response. On the other hand, when I strengthen a connection by removing something she doesn’t want, I’m negatively reinforcing the behavior.
Punishment involves introducing something unpleasant to make a particular behavior less likely. Sternly telling Gemini that she’s not allowed to eat the cat food is my attempt to punish her, introducing something unpleasant (disappointed dad voice) to make a behavior less likely (eating her sisters’ food). This hasn’t worked yet – maybe I’m not doing the voice right – but my point is that punishment isn’t negative reinforcement. Reinforcement is about making a behavior more likely; punishment is about making a behavior less likely.
This is a nice example of how technical terminology can be subtly misleading when we’re not careful.
Some terms inherit connotative baggage from their constituent parts. In this case, because good things are “positive” and bad things are “negative,” it’s easy to assume that positive reinforcement involves reward and negative involves punishment. In and of itself, there’s nothing wrong with this, but it’s inconvenient for psychology teachers, since students have a hard time unlearning the misleading connotation. I could make the same point about the terms “validity” and “soundness” in the context of a logic class. Both terms have rich colloquial meanings that bear little resemblance to the technical notion, which means that logic professors have to remind their students often that “valid” doesn’t mean “true” or “reasonable,” not in logic class anyway.
These are just obstacles to introducing technical terms, or academic jargon, to people unfamiliar with them. No big deal. But if you know about the risks involved with connotative baggage, then it makes sense to take care when using language creatively – drawing an analogy, trying out a metaphor, or coining a new term for a concept whose contours you’re still working out. (“Connotative baggage,” for example.) If you know there’s a rake in the lawn, doesn’t it make sense to look out so you don’t step on it?
In recent years, I’ve seen a lot of excitement in the philosophical community around notions that have bubbled up in public conversation. “Fake news” and “gaslighting” come to mind, but so does “conspiracy theory.” These terms have some things in common. They are richly textured and informal, requiring creativity from speakers as well as interpretive flexibility from the audience. They are also normatively thick and value-laden, rather than being flatly descriptive. It’s hard not to hear an accusatory tone every time you hear them used, because it’s built into the meanings of the words.
Anyone who wanted to theorize about the nature of fake news, gaslighting, or conspiracy theories – or just wanted to engage in some light conceptual analysis – would have to contend with some unwieldy connotative baggage. This is an especially acute problem for “fake news”. There’s a sense in which we were all there for its coinage and christening, and hence that its meaning must be publicly available and uncontroversial, but I suspect that this is a false sense of security. Your peers might let you get away with using these words in lots of creative ways, but that doesn’t mean that their meanings are clearly and mutually understood. Lots of peer groups have been using this very phrase in lots of ways since 2016, and what flies in one might not in another.
I’m not trying to tell anyone what they’re allowed to write or think about. But I do think that the problem of connotative baggage means that philosophizing about something like fake news should not be confined to coming up with a theory about it. Sometimes the best we can do with an interesting idea or loaded term is to make shallow generalizations about it and go from there. (Interviewing Iris Murdoch, Bryan Magee asked the great philosopher-cum-novelist, “How would you characterize literature as distinct from philosophy?” She answered, “I don’t think one can offer exactly a definition of ‘literature’ nor is it necessary in place here. We know what literature is. It’s an art form using words.”)
Set aside the question of what point there is to philosophizing about conspiracy theories – I want to ask what we could hope to achieve if we ever had a satisfactory definition or account. What comes next? I have the uneasy feeling that, no matter how fastidious we are in having a theory of a term like that, we will still be left with the problem of connotative baggage. I ran into this problem hard when teaching my students a piece on the “epistemology of conspiracy theories.” I thought the piece had been reasonably clear about what the subject matter was, but that was a mistake on my part. Most of the discussion circled around clarifying what the author wasn’t talking about (not, to my students’ disappointment, which conspiracy theories are true), rather than whether he was right about what he was talking about.
It gave me flashbacks to teaching logic and trying to get clear on logical validity. But “logical validity” is a piece of proudly technical jargon. It inherits its connotative baggage from colloquial English as a matter of historical accident – the word that logicians use just happens to sound the same as a colloquialism. However, it seems like a mistake to try to make technical jargon out of a recent colloquialism, by appropriating the colloquialism and sanding off the rough edges. First there’s the inevitable throat-clearing: “I’m going to talk about X, but let me be clear that I’m not talking about X in the way you do.” But there’s also the illusion that, having cleared our throats, we’re free to use the term that we’ve domesticated without any confusion. With all the dynamism of colloquial usage cleared away, we can say what we really wanted to say the whole time.
I have my doubts, but I think there’s more to worry about than pedagogical and communicative obstacles. A more polished product might leave us with less to say, less expressive ability, rather than more. With some of our colloquialisms, the lumbering connotative baggage is a feature, not a bug, partly because it enables the flexibility that we want out of those phrases. Once we turn “conspiracy theory” into a precision instrument meant strictly to refer to epistemically defective just-so stories involving small groups of colluding parties, it leaves us unable to talk seriously about actual cases of small groups of colluding parties getting up to no good, say, by burying inconvenient findings about the effects of carbon dioxide, or breaking into a political rival’s campaign offices. These are conspiracies – we wouldn’t want to make it harder to talk about them as such. But we risk doing just such a thing by throwing away connotative baggage in the name of clarity.
This doesn’t mean we have to give up our precision tools. My plea is for fewer merely verbal disputes in public conversation. If there’s a philosophically juicy feature of a phenomenon that could reasonably be flagged as gaslighting (for example), it might be more worthwhile to articulate what’s interesting there, without insisting on isolating that feature as an essential feature of gaslighting or a newly discovered subspecies. We can have our precision and clarity without sacrificing connotative baggage.