Fake News and Signal Detection Theory
In which we separate truthful signals from fake noise
I.
Our brains are flooded by sensory information every waking second. Without ways to filter all that incoming information, we’d go crazy. But we can’t ignore everything, of course. In that endless sea of sensory noise, there are signals.
How do we separate signal from noise?
Our story begins in the mid-19th century, when German psychologist Gustav Fechner - a pioneer in experimental psychology as well as a physicist - asked that same question. He was fascinated by how people detect weak sensory signals in the presence of noise. In Fechner’s original experiments, people were asked to hold up two weights and determine which was heavier. Fechner found that people could detect weak signals (differences in weight) in the presence of noise (the feeling of the material, the researcher giving instructions, and so on).
Fechner unleashed his mathematics on it and signal detection theory was born. Signal detection theory provides a framework for understanding how we separate signal from noise and how factors such as signal strength, noise level in the environment, and perceptual biases can influence this.
In the late 1920s, American psychologist Louis Leon Thurstone ran similar experiments with a twist: he also asked about subjective feelings (for example, “Which artwork generates the strongest feelings of beauty?”).
A few decades later, in the 1950s, other researchers added three crucial observations:
Fechner’s signal detection theory applies in many contexts.
A lot of the noise in our heads can be sampled by conscious awareness.
Different people have different receiver operating characteristic (or ROC) curves.
Translation for point three: there is variability in how well people can pick ‘true’ signals from noise distributions. These differences can be affected by many factors, such as age, gender, personality, and cognitive abilities.
II.
Signal detection theory has been used to study cognitive processes including perception, attention, memory, and decision-making. After all, the signals - regardless of whether they’re real or fake - that are selected to become part of our conscious awareness color how we see and experience the world.
If we extend that application of signal detection theory beyond the terrible and boring tyranny of normalcy’s benchmarks, it might, perhaps, be an interesting framework to look at certain types of neurodivergence? Good call. I got you.
Applying signal detection theory to cognitive mechanisms at play in mood and anxiety disorders suggests that people with these conditions have an affective bias. Another way to see this is as an increased tendency of anxious/depressed individuals to predict lower rewards resulting from the decisions they make.
Another example is the occurrence of auditory hallucinations, for example, in schizophrenia. These hallucinations are false positives, noise interpreted as a signal in a kind of ‘hypervigilance’. Vice versa, people with ADHD might have trouble finding signals in a noisy environment (a ‘vigilance deficit’ which makes it harder to ‘pay attention’), although this is quite context-dependent.
Also, in uncertain situations, people with autism are faster to act as though something is a signal, even if their sensitivity to signals is the same as for people without diagnosed autism. (Although there are studies that find a higher sensitivity for, for example, pitch, in individual switch autism. Context matters, again.) Finally, some people with borderline personality disorder are less sensitive to pain stimuli, which might correlate with a tendency for self-injury.
These differences in signal detection need not be constrained to diagnosed neurodivergence, though.
Here’s an example I couldn’t resist sharing: people who score high on extroversion are more likely to be fooled by phishing attempts. In other words, the persuasion principles used in phishing emails are more likely to convince extroverts that they are signals rather than noise.
That’s interesting because phishing is only one of the relatively new sources of virtual noise we are exposed to in our modern online lives. We are flooded by online content, tailored for engagement rather than for accuracy or nuance. The examples above show that people differ in how sensitive they are when it comes to separating signal from noise, and that this is context-dependent.
Can we use this to vaccinate ourselves against fake news?
III.
Is it just me, or does the world feel increasingly noisy?
Clickbait, phishing, fake news. It all adds up to a cacophony of noise that clamors for our attention. The Internet and social media provide a conduit for spreading information widely and rapidly regardless of its truthfulness. Even if false information is later refuted, it can still continue to affect people's judgments and decisions by introducing extra noise that makes it harder to find true signals.
Very roughly, four non-mutually-exclusive reasons have been proposed for why people (don’t) fall for fake news.
Partisan bias: we’re more likely to think that information that aligns with our personal beliefs is true.
Cognitive reflection: the likeliness of identifying fake news is associated with the ability for cognitive reflection (the ability to override an incorrect gut response).
Motivated reflection: a combination of the two previous ones - using your cognitive skills to rationalize the news so that it is consistent with beliefs you want to ‘protect’ (AKA posthoc rationalization).
Prior exposure: we are more likely to accept information that is more fluent to process, which, in turn, is driven in part by prior exposure - social media is repetitive because that is what works.
A 2021 study used a signal detection approach to figure out which one(s) of those play(s) the biggest role in how we (don’t) fall for the false claims in our feeds.
As you’d expect, ideological beliefs influence judgments through biases. We accept ideology-congruent news as real and dismiss ideology-incongruent news as fake.
Cognitive reflection increases the accuracy in discriminating real from fake news and makes it more likely to judge news as fake. It does not, however, save you from your own ideological biases, which suggests that motivated reflection may not be that important.
Prior exposure reduces discrimination ability and increases the likelihood of judging news as real, regardless of whether it’s true or not.
The authors point out that accurately assessing news hinges upon:
… the capacity to disentangle two conceptually distinct aspects in the identification of fake news: (a) ability to correctly distinguish between real news and fake news and (b) response biases to judge news as real or fake regardless of news veracity.
Think you’ve got what it takes to sniff out fake news? Here you can take the University of Cambridge's ‘Misinformation Susceptibility Test‘.
Let me know how you did. Here’s me:
Related thoughts:
Phew, this one took a while to research; let me know if it was too long (thanks for reading to the end). And if you enjoyed it, you know the drill: share, like, subscribe, etc.
Excellent article! Add the periaqueductal gray (PAG) area of the brain stem (which is the hub for self-awareness and pain mitigation) to the Synaptic Adhesion Molecules (SAMs) (which cement our thinking to the local electrical minima) and add to these our ease seeking tendencies and you have the perfect recipe for migration to the lowest common denominator of thinking.