Beyond Certainty: What an Assamese Tiger Story Reveals About How We Understand the World

Most people encounter Bayesian reasoning through medicine.

A doctor explains that a “99 per cent accurate” test does not necessarily mean there is a 99 per cent chance of disease. The final probability depends not merely on the laboratory machine, but also on how common the disease is in the population being tested.

At first glance, this feels absurd.

If the machine is highly accurate, why should prevalence matter? Surely truth is already inside the test itself.

Yet this strange idea opens the door into one of the deepest ways of understanding reality. Because Bayesian reasoning is not really about statistics alone. It is about how humans update belief in an uncertain world. And once one notices it, Bayesian logic appears everywhere — in medicine, politics, social trust, media ecosystems, economics, artificial intelligence, and even Assamese folk wisdom.

Long before mathematics formally described it, human societies were already living Bayesianly.

There is an old Assamese proverb: “এশ গৰু মাৰিলে বাঘৰো মৰণ” — after killing a hundred cattle, even the tiger eventually dies.

 

At one level, it is a simple rural observation. A tiger repeatedly preying on village cattle eventually attracts vigilance, organised response, and retaliation. But beneath the folk imagery lies something much deeper.

The proverb quietly encodes a probabilistic understanding of the world.

One missing cow proves little. It may have wandered away, been stolen, or died naturally. But as losses accumulate, villagers unconsciously update their understanding of hidden reality. After enough incidents, uncertainty collapses into near-certainty: there is indeed a tiger nearby.

 

The villagers’ mental model changes continuously with accumulating evidence.

That is Bayesian reasoning in folk form.

The proverb contains another subtle insight too. The tiger’s repeated success eventually becomes self-destructive. Each successful attack increases the probability of exposure. Patterns accumulate until hidden causes reveal themselves.

 

Long before modern statistics, agrarian societies had already learned that evidence acquires meaning through repetition, context, and prior experience.

The formal mathematical roots of this idea trace back to Thomas Bayes, an unlikely figure whose name eventually became attached to one of humanity’s most important inferential frameworks.

 

Bayes himself was not attempting to revolutionise medicine or artificial intelligence. His original work was modest and unpublished during his lifetime. But the question he asked was transformative.

Before Bayes, probability theory mostly worked in the “forward” direction: if we know the probabilities, what outcomes should we expect?

Bayes asked the reverse: if we observe outcomes, what can we infer about hidden reality?

That reversal changed everything.

 

Because almost all human reasoning works this way.

Doctors infer disease from symptoms. Scientists infer laws from experiments. Intelligence agencies infer hostile intent from fragmentary signals. Citizens infer political credibility from repeated behaviour. Humans rarely observe truth directly. They infer it probabilistically from incomplete evidence.

The person who truly recognised the enormous implications of this framework was Pierre-Simon Laplace, who independently rediscovered and vastly expanded Bayesian reasoning into a broader theory of rational inference.

 

Yet history preserved Bayes’ name.

Ironically, Bayesian reasoning itself helps explain why. Once early scholars attached the theorem to Bayes, generations of textbooks, institutions, and academic memory inherited that association and reinforced it continuously over centuries. Even Bayes’ legacy evolved Bayesianly.

The most famous modern illustration comes from medicine.

 

Suppose a disease affects only 1 per cent of a population. Now imagine a diagnostic test claiming 99 per cent sensitivity and 99 per cent specificity.

Most people instinctively assume that a positive result means a 99 per cent chance of disease.

But the arithmetic tells a different story.

 

Imagine testing 10,000 people. Only 100 truly have the disease. The test correctly identifies 99 of them. However, among the remaining 9,900 healthy individuals, 1 per cent will still test positive falsely — producing another 99 positive results.

The outcome becomes startling: 99 true positives and 99 false positives.

A person testing positive therefore has only about a 50 per cent probability of actually having the disease.

 

The machine did not fail. The chemistry did not fail. The context changed.

This is the central Bayesian insight: evidence never possesses fixed meaning independent of background reality.

The same signal means different things in different environments.

 

Medicine, in truth, is Bayesian at its core. Clinicians constantly integrate: age, epidemiology, travel history, symptom pattern, local prevalence, prior probability, and evolving evidence.

A laboratory result never exists in isolation.

Politics and public life increasingly operate similarly.

 

Imagine a dramatic video clip appearing online showing a politician making an inflammatory remark. If the politician already possesses a history of such rhetoric, the public quickly accepts the clip as genuine. But if the same clip depicts a measured public figure with decades of disciplined conduct, many immediately suspect editing or selective context.

 

The evidence remains identical. The interpretation changes because prior probability differs.

 

Social media ecosystems constantly exploit this weakness in human cognition. Humans naturally overweight vivid anecdotal evidence while neglecting statistical base rates. Rare crimes create nationwide panic. Isolated incidents shape perceptions of entire communities. Repeated narratives gradually alter public priors until even ambiguous future evidence becomes interpreted through accumulated expectation.

 

The Assamese proverb about the tiger captures this process elegantly. Patterns reshape belief landscapes.

In the twentieth century, statistics briefly moved away from Bayesian thinking. Classical “frequentist” statistics attempted to eliminate subjective prior belief and focus only on repeated measurable outcomes. This produced powerful tools like p-values and confidence intervals.

But reality proved more complicated than rigid frameworks alone could comfortably handle.

 

As computing power expanded, Bayesian methods returned dramatically. Today Bayesian reasoning underlies: machine learning, weather forecasting, spam filters, recommendation algorithms, financial modelling, genomic prediction, and artificial intelligence itself.

Modern AI is deeply Bayesian in spirit. It continuously updates probabilities as new information arrives.

 

In effect, Bayesianism became the mathematics of learning.

At its deepest level, Bayesian reasoning carries an unusual intellectual humility. It assumes: certainty is rare, evidence is contextual, beliefs should evolve continuously, and no observation exists entirely outside prior assumptions.

This stands in contrast to ideological thinking, which often treats isolated evidence as final proof detached from broader statistical reality.

A Bayesian mind neither clings rigidly to certainty nor collapses into total scepticism.

Instead, it updates.

Continuously. Probabilistically. Contextually.

Perhaps that is why Bayesian reasoning feels so profound once one truly grasps it.

It reveals that reality is not merely made of isolated facts. Facts derive meaning from surrounding context, accumulated history, prior expectation, and evolving uncertainty.

The old Assamese villagers watching cattle disappear one by one already understood something modern civilisation often forgets: patterns matter more than isolated events.

The tiger was never identified with certainty from a single missing cow. Knowledge emerged gradually through accumulated evidence interacting with lived experience.

And in an age saturated with algorithms, propaganda, diagnostics, social media outrage, and competing realities, that way of thinking may become not merely useful, but essential.

 

(Author’s Note: Dr Jayanta Biswa Sarma writes on politics, institutions, and society through the lenses of history, philosophy, and systems thinking, drawing on both Indian and Western intellectual traditions. Artificial intelligence tools may be used in preparing this article as research and editorial aids. All arguments, interpretations, and final editorial judgement remain the author’s responsibility)