Myths and legends about monsters have excited the human imagination for hundreds of years. Although vampires, werewolves and ghosts do not exist in reality, there are irrational belief constructs that are equally monstrous. Not just in content, but also in consequence. These are often based on exploiting common human tendencies with an additional layer of motivated reasoning reinforced by pseudoscience. This article will examine one such monster known as the the “poisonous M&Ms analogy”. It is often deployed as a way to prop up indefensible stereotypes by taking advantage of human ignorance about base rates, risk assessment and criminology. In the end, it tries to divert attention from the inherent bigotry in making flawed generalizations.
What is the poisonous M&Ms analogy?
The typical deployment of the poisonous M&Ms analogy goes something like this:
You say that I am overgeneralizing about [group X]?
Imagine a bowl of M&Ms. 10% of them are poisoned. Go ahead, eat a handful of them. After all, they are not all poisonous!
The general idea is that the overgeneralizing about [group X] and treating them all as suspect is just as valid as not wanting to eat an M&M from a bowl were some of them are poisonous. This may seem reasonable at first, but it quickly becomes apparent that this is merely an intuition pump designed to automatically dismiss criticisms about flawed generalizations without due consideration.
Follow Debunking Denialism on Facebook or Twitter for new updates.
Why is the poisonous M&Ms analogy monstrous?
Because it can be used to prop up any kind of harmful stereotype about groups such genders, ethnicities, religious and political communities without having to engage the objections to unfair generalizations. In reality, the poisonous M&Ms analogy is a more manipulative version of the general tactic known as the “I know not all X are Y, but [flawed generalization]”. Here are some examples from the white supremacist website Stormfront:
So, it’s true that not all blacks are criminals, but apparently, all (or at least, most) blacks support criminal behavior.
So let me state again not all blacks are criminals but they are more prone to criminal behavior than any other race.
Not all blacks are criminals but they are overrepresented in interracial attack rates amongst other serious crime rates.
These flawed claims can easily be debunked by noting that the vast majority of African-Americans are not criminals, any overrepresented is small and not due to their ethnic status itself.
What sets the “Poisonous M&Ms” formation apart is that it is tries to defend discriminatory stereotypes by pumping intuitions in people who are statistically illiterate rather than to promote overt absurdities that most people already know are erroneous.
Why is the poisonous M&Ms analogy flawed?
There are many reasons why the poisonous M&Ms analogy does not work. Here is a short survey of some of the more egregious flaws:
Little to no specificity: because the argument has essentially no specificity, we can revert the argument back to the group making it. If white supremacists use it to support their indefensible stereotype of African-Americans as criminals, we can apply it back to white supremacists. If conservatives make the argument against liberals, the argument can be sent back with the corresponding stereotype of conservatives. Here is how it would look when it is reverted back against white supremacists: “You say that I am overgeneralizing about white supremacists being criminals? Imagine a bowl of M&Ms. 10% of them are poisoned. Go ahead, eat a handful of them. After all, they are not all poisonous!” No white supremacist would accept that argument as reasonable, which means they cannot reasonably deploy it against ethnic minorities either.
Base rate neglect: a rational risk analysis must take base rates into account, not just the consequences. Even if the consequences of an event is large and negative, the probability of the event might be low. The “poisonous M&Ms analogy” does state a non-empirical base rate (10%), but it never figures in the risk evaluation because of the way the analogy is constructed. It tries to make the case that you should not eat M&Ms even if the base rate of getting a poisonous one is small. In fact, the analogy does not just neglect base rates, it ignores its relevance completely.
Assumes that “risk-free” is possible: the analogy also tries to exploit the human tendency to think that it is possible for an event to be risk-free. After all, the moral of the analogy is that even if there is a small risk of getting poisoned, it is reasonable to not take one. You only want to eat M&Ms if there is virtually no risk of getting poisoned, right? Most people would probably not eat one of the M&Ms even if there was a 1% or a 0.1% chance of being poisoned. In reality, any event such as walking across the street, traveling in a car or drinking a glass of water is not risk-free. However, proponents of this analogy would never argue that you should not drink water because of a small risk of choking.
Not poisonous to you: even if you happen to come across an individual from group X that fits with the stereotype does not mean that you are in danger. For instance, most crimes are committed by a small number of criminals and so most criminals commit very few crimes. Thus, even if the person of group X you come across happens to be a criminal, it does not automatically mean that you are in any particular danger. However, in the analogy, the poisonous M&Ms are obviously poisonous to humans in general.
Not a random sample: base rates apply to a random sample. Your friends, colleagues, dates or people you walk past in the night do not constitute a random sample from the underlying population. So even if we accept a given base rate (such as 10% in the analogy) does not mean that a given person of group X has a 10% probability of being a criminal. However, the analogy want us to believe that a given M&M has a 10% risk of being poisonous. In reality, a bowl of M&Ms usually come from the same bag or batch, so even that does not qualify as a random sample. So the argument essentially tries to compare two non-random samples and say the two situations are analogous.
Predictors exists: it is commonly believed that you cannot tell criminals apart from non-criminals. However, this is not true as there exists several predictors of criminal behavior: childhood maltreatment, failing school, poor moral reasoning and empathy, excessive alcohol and drug use, certain personality traits such as impulsivity and insensitivity as well as hanging out with criminals and extreme commitment to traditional masculinity (Bernard, Snipes and Gerould, 2010, p. 353). However, the analogy assumes that all M&Ms look the same whether or not they are poisonous or not. If there was a way to distinguish the two, it would not matter that a certain proportion are toxic as you could just not pick them.
Conclusions
The Poisonous M&Ms analogy is an irrational monstrosity as it can be deployed in an effort to dismiss any criticisms of flawed stereotypes. However, it has numerous flaws: it can be used against any group without specificity, it ignores base rates, it assumes that risk-free existence is possible, it fails to grasp the important of a random sample and promotes several false claims about criminology. In the end, the poisonous M&Ms analogy is just a dishonest tactic to whitewashing bigotry.
References
Bernard, Thomas J., Snipes, Jeffrey B., & Gerould, Alexander L. (2010). Vold’s Theoretical Criminology (International Sixth ed.). New York: Oxford University Press.