Beliefs as Probabilities

The idea that strong, unchanging convictions are always a virtue is a myth. Treating beliefs as probabilities means seeing them not as simple true/false statements, but as a level of confidence between 0% and 100%. For example, instead of thinking "This project will succeed," you might think, "I'm 85% confident this project will succeed given our current resources." This simple shift keeps you open to new information and helps you adjust your views without feeling like you were "wrong."

Reflect on this: What's one strong opinion you hold, and what would it take to shift your confidence by just 10%?

Thinking of beliefs as probabilities is a core tool for rational decision-making. It replaces the rigid, binary world of true/false with a flexible spectrum of confidence. This mental model encourages you to see your beliefs as hypotheses to be tested, not as identities to be defended.

When a belief is held at 100% certainty, we instinctively engage in confirmation bias, seeking only evidence that supports our view. But if you hold a belief with 90% confidence, you implicitly acknowledge a 10% chance you might be wrong. This small gap of uncertainty keeps you curious and open to evidence that could challenge your perspective, leading to more accurate worldviews over time.

This approach has powerful real-world applications. A doctor doesn't think, "This patient has the flu." Instead, she might think, "There's an 80% probability it's the flu, a 15% chance it's a common cold, and a 5% chance of something else." This allows her to order the right tests to update her probabilities and arrive at a more accurate diagnosis.

Similarly, an intelligence analyst tracking a global conflict doesn't deal in certainties. They provide policymakers with probabilistic assessments, like "There is a 65% chance that negotiations will break down within the next month." This communicates the level of uncertainty and helps leaders make better-calibrated decisions, avoiding the traps of overconfidence or unwarranted fear.

This concept is a cornerstone of Bayesian epistemology, a branch of philosophy and statistics that formalizes how a rational agent should update their confidence in a proposition when they encounter new evidence. Rather than seeing beliefs as static, Bayesian reasoning treats them as 'degrees of belief' or subjective probabilities that are continuously refined through a process known as Bayesian updating.

• Jaynes (2003): Argues that probability theory is not just a tool for analyzing random events like coin flips, but a fundamental extension of logic itself, allowing us to reason consistently in the presence of uncertainty. [1]
• Duke (2018): Translates this concept for practical decision-making, framing decisions as 'bets' on an uncertain future. By thinking probabilistically, we can separate the quality of a decision from the quality of its outcome, leading to better long-term results. [2]
• Tetlock & Gardner (2015): Through the Good Judgment Project, they found that the most accurate forecasters ('superforecasters') are distinguished not by what they believe, but how they think. They constantly update their probabilistic beliefs in small increments in response to new information. [3]

While powerful, thinking in probabilities isn't a cognitive cure-all.

• Cognitive Overhead: It is mentally demanding to assign precise probabilities to all our beliefs. For low-stakes, everyday decisions, heuristics (mental shortcuts) are more efficient.
• The Reference Class Problem: The probability you assign can depend heavily on the reference class you choose. Is a specific medical procedure's success rate based on all patients, patients of your age, or patients with your exact health profile? The choice dramatically changes the probability.
• Poor Intuition: Humans are notoriously bad at intuitively understanding and manipulating probabilities, often falling prey to biases like base rate neglect.