The Illusion of Perfect Reasoning

Why our decision-making often relies on flawed arguments

A previous version of this post was published on Psychology Today.

Source: Pixabay

The defendant didn’t tell his side of the story after he was arrested.

Innocent people never invoke their Fifth Amendment right to remain silent.

Therefore, the defendant is guilty¹.

Arguments like this often appear in discussions around high-profile cases and on social media, where people are quick to draw conclusions about guilt or innocence. But what exactly is an argument? At its core, an argument is a series of statements (premises) that allows us to infer a conclusion (or claim)². While arguments like the one above might seem convincing on the surface, they are often flawed. To understand why, it's helpful to distinguish between the two main types of arguments: deductive and inductive³.

Deductive Reasoning

The example presented at the beginning was a form of deductive reasoning. In deductive reasoning, the inference from the premises to the conclusion is airtight—if the premises are true, then the conclusion must be true. Such arguments are considered logically valid. Here’s an example:

All healthy trees have leaves in the summer.

An oak is a type of tree.

Therefore, a healthy oak tree has leaves in the summer.

If both premises are true, then the conclusion must be true. This makes the argument logically valid, regardless of whether the premises themselves are actually true.

Now, let’s consider an alternative argument:

An oak is a type of tree.

A healthy oak has leaves in the summer.

Therefore, all healthy trees have leaves in the summer.

In this case, even if the premises are true, the conclusion doesn’t logically follow. We cannot infer from the properties of a single oak tree that all trees must have leaves in the summer. Therefore, this argument lacks validity.

The validity of the argument, though, only describes whether the conclusion logically follows if the premises are true. Premises aren’t always true. In such cases, no matter how valid the deductive argument is, the argument would be said to be unsound.

Let’s return to the guilt or innocence argument at the beginning:

The defendant didn’t tell his side of the story after he was arrested.

Innocent people never invoke their Fifth Amendment right to remain silent.

Therefore, the defendant is guilty.

This argument is logically valid if we assume both premises are true. But a problem lies in the second premise—it isn’t true that innocent people never invoke their 5th Amendment right to remain silent. People choose to remain silent for many reasons, including that lawyers will often advise their clients to exercise their fifth amendment rights regardless of their guilt or innocence. Because the second premise lacks truth, the argument becomes unsound as a deductive argument⁴.

To explore an alternative approach, we can restructure our argument so that it is inductive rather than deductive. This allows us to focus on the likelihood of a conclusion based on available evidence, rather than trying to claim certainty.

Inductive Reasoning

In inductive reasoning, the goal is not to guarantee that the conclusion must be true, but rather to provide a probable conclusion based on the available evidence⁵. Unlike deductive reasoning, where the conclusions are supposed to follow with certainty if the premises are true, inductive reasoning allows for degrees of uncertainty.

Let's revisit the earlier example, but modify it to illustrate inductive reasoning:

The defendant didn’t tell his side of the story after he was arrested.

Innocent people seldom invoke their 5th Amendment rights to remain silent.

Therefore, the defendant is probably guilty.

This argument no longer presents an airtight conclusion like deductive reasoning does. Instead, the argument suggests a probable conclusion—that the defendant is guilty—based on the premises⁶. The strength of an inductive argument depends on how strongly the premises suggest the conclusion, rather than guaranteeing it. Inductive reasoning conclusions are inherently uncertain, and the strength of an argument reflects how probable it is that the premises lead to the conclusion. However, an inductive argument can only be considered cogent if it is both strong and based on true or well-supported premises. If the premises are weak or unsubstantiated, the argument lacks cogence, meaning our confidence in the conclusion should be proportionally lower.

To illustrate using the example above, it may appear to be a strong argument at first glance, but there’s a major flaw. The second premise—that “innocent people seldom invoke their Fifth Amendment right to remain silent”—may not be sufficiently supported by evidence. There are many valid reasons why an innocent person might remain silent, such as legal advice, a fear of misinterpretation, or simply a desire to avoid self-incrimination⁷. And so, without sufficient evidence to support the second premise⁸, the claim or implication that such a conclusion is probable lack cogence, which requires both strong premises and true evidence.

Inductive reasoning is more open-ended and flexible than deductive reasoning, making its use extremely common in everyday life. We often generalize from specific experiences or observations to make predictions about the future. For example, if you've seen that a particular restaurant has had long waiting times on multiple occasions, you may conclude that it will probably have a long wait time the next time you visit. However, such conclusions are not guaranteed—you could go there one day and find it completely empty. The inductive argument is strong but not infallible, and that distinction is important.

Inductive reasoning can take many forms, including generalizations, causal inferences, and analogies. Regardless of the type, it's essential to consider both the strength of the evidence and the reasoning behind the conclusion. A strong inductive argument provides a high likelihood that the conclusion is true, but it does not guarantee it—there's always the possibility that new evidence could lead to a different conclusion.

Where We Make Errors

Although we all reason, we often fail to pay sufficient attention to the strength of our reasoning or the quality of the evidence underlying it. There are several ways this can impact our decision-making, and I want to emphasize four in particular.

Believing We’re Arguing Deductively When We’re Really Arguing Inductively. As evidenced in the initial example for this post, many people believe their argument is airtight. This can lead people to jump to conclusions on the basis of rather meager evidence. Consider, for example, the reasoning that leads to the illusion of causality. A person might think, "Every time I wear my lucky socks, my team wins. Therefore, my socks must be the reason for their success."

This argument is actually inductive—it draws from specific observations to propose what, at best, would be a probable conclusion rather than a guaranteed one. Yet people often believe that such arguments are logically airtight, treating them as if they were deductive conclusions. This can lead to overconfidence in one’s conclusions, which may be harmless when it comes to lucky socks but can have far more serious implications in other decision-making contexts, such as health decisions or financial investments.

The confounding of deductive with inductive argumentation can be even more pronounced when we hold strong biases that lead us to want to draw certain conclusions. In these instances, confirmation bias comes into play—we seek out information that supports our existing beliefs while ignoring or dismissing contradictory evidence. In the case of the lucky socks, a fan might selectively remember all the times their team won when they wore the socks but disregard the losses⁹. Alternatively, if we return to our original example, if we already decided the defendant is likely guilty, we may take his silence as evidence that merely reinforces that conclusion. This tendency to confirm what we already believe further reinforces the illusion of deductive certainty, making us even more confident in conclusions that are actually just probabilistic.

Failing to Identify a Probability Standard. Even if we recognize that our argument is inductive, we often fail to adequately consider how probable our conclusion is—and more importantly, how that probability should influence our actions. Not all probable conclusions are equally convincing or reliable. For instance, concluding that an investment will probably yield returns doesn’t mean you should make that investment without further scrutiny. The strength of that conclusion might vary depending on whether it's 85% probable or 60% probable, especially if making the investment involves significant risk.

The problem is that we sometimes treat all probable conclusions as if they have similar weight. However, an argument that is 51% likely to be true (i.e., just more likely than not) is very different from one that has an 85% likelihood. In some cases, an argument that barely crosses the threshold of being probable may not provide sufficient confidence to justify acting on it, particularly in high-risk scenarios.

For example, suppose a doctor concludes that a treatment is "probably effective" for a patient based on available studies. This could, of course, mean somewhere between 51% and 100% probable. At a very high level of probability, say 85%, the treatment might be worth pursuing, assuming the potential benefits of successful treatment are worth the possible side effects that come with it.

However, as the probability gets lower, closer to 51%, the decision becomes far more complex. The risks, costs, and potential side effects of the treatment may no longer be justifiable when the likelihood of success is only slightly better than a coin flip. In these cases, both the doctor and patient might hesitate or even decide against the treatment. This highlights why simply labeling something as "probably effective" is insufficient; being cognizant of different degrees of probable matters greatly in determining whether action is warranted, particularly in high-stakes scenarios like medical treatment. In such situations, it may be helpful to assess whether a given desired outcome is probable enough.

Overconfidence can also contribute to this misjudgment. We often have an inflated sense of our ability to accurately assess probabilities, leading us to overestimate how likely our conclusions are to be true. This overconfidence can make us more prone to taking risks without properly evaluating the level of uncertainty involved. For example, in the case of the "probably effective" treatment, if the doctor or patient overestimates the likelihood of success, they may proceed despite the potential risks outweighing the benefits. Recognizing the potential role of overconfidence is important for appropriately assessing the probability of outcomes, especially when those outcomes carry significant risks.

Committing Logical Fallacies. When we reason, we apply logic to make inferences based on new or existing information. The strength of those inferences depends on the validity or strength of the logic and the quality of information we use. But our brains are wired to draw conclusions based on incomplete information, and this can cause us to make erroneous inferential leaps that invalidate or weaken our argument. These reasoning errors are known as logical fallacies.

For example, we might fall into the trap of affirming the consequent, which is a logical fallacy. Consider the following deductive argument:

• If it rains, the street will be wet.

• The street is wet.

• Therefore, it must have rained.

This fallacy occurs because we're drawing a conclusion based on an observed effect (the wet street). We ignore the possibility of alternative causes—a hose, a water main break, or some other cause. This selective attention to information, a hallmark of heuristics, simplifies our reasoning but can result in flawed conclusions due to the oversimplification of causation. The mental shortcut of assuming a single cause leads us to affirm the consequent, making the argument logically invalid from a deductive standpoint.

We could, however, restate our claim as one of probability—it probably rained— which would make the argument inductive. In this case, if rain is the most common reason for a wet road, then it's a reasonable inductive conclusion to think it rained, as long as we're comfortable with the possibility of some uncertainty in the conclusion.

Fallacious reasoning renders a deductive argument invalid (and thus unsound), but it also weakens inductive arguments, potentially to the point of being considered fatally flawed. For example, suppose you meet two people from a particular city who are rude, and you then conclude that everyone from that city is likely rude. This is an inductive argument because it’s based on specific observations used to draw a conclusion, but the reasoning is flawed because the sample size is too small to support the conclusion you reached. Therefore, the argument lacks strength because there is insufficient evidence on which to draw the conclusion you did.

Failing to Consider the Credibility of Evidence Underlying the Premises. It’s quite easy to look for evidence that supports the premises we use to make an argument, a tendency known as confirmation bias. When we focus solely on evidence that aligns with our beliefs, we neglect to evaluate whether the evidence itself is credible or representative. This oversight can result in arguments that are easily refuted or fundamentally flawed. The soundness or cogence of an argument ultimately depends not only on logic underlying it but also on the quality of the evidence that supports the premises.

For example, suppose someone claims a specific diet works because a few people they know tried it and lost weight. The evidentiary basis for this inference is weak. We don’t know whether those people are representative of the general population or whether other factors might have contributed to their success. As a deductive argument, it would be unsound because the premises do not provide sufficient and evidence to guarantee the truth of the conclusion. As an inductive argument, it would lack cogence because the sample size is too small and unrepresentative—the evidence is not strong or reliable enough to support a probable conclusion that the diet would work.

When we fail to scrutinize the credibility of evidence, we often end up overestimating our confidence in a conclusion. Adjusting the level of confidence in a conclusion so it matches the quality and credibility of the evidence becomes extremely important in scenarios with significant consequences. By critically evaluating the evidence we use, we can avoid making leaping to conclusions based on weak or misleading information.

Conclusion

There are no magic formulas for avoiding the pitfalls of flawed reasoning, nor a one-size-fits-all answer to how serious each trap might be. Every situation is unique, and the consequences of relying on flawed reasoning can vary widely. In high-stakes scenarios, more cogent or sound arguments become extremely important, while situations with fewer consequences may afford more leniency.

To improve decision making, it helps to better understand our own thought processes, both in how we consciously and unconsciously formulate arguments—known as metacognition. By cultivating awareness of our reasoning patterns and taking the time to question our assumptions, we can make more informed choices and minimize the influence of fallacious reasoning.

The first step towards better reasoning is recognizing that everyone, at some point, relies on flawed logic. The goal isn’t to become infallible, but rather to consistently question the reasoning behind our decisions, remain open to new evidence, and develop habits of reflective thinking. These actions can ultimately lead to more reliable and effective decision making, helping us navigate a complex world with greater insight and confidence.

1. This is not a direct quote, but it is an insinuation made often by the general public, especially on social media outlets. We can rework the premises in any number of ways to still derive a valid deductive argument. For example, we could change the second premise to “Only those who are guilty invoke their 5th Amendment rights.”

2. We can contrast this with the way many people use the word in common parlance, which often is more synonymous with a heated disagreement with a strong emotional component.

3. A third type, abductive, is also sometimes discussed. It appears to be more a form of inductive reasoning focused on sensemaking in the absence of strong premises (i.e., a best guess). Given the lack of agreement surrounding this topic, it’s well beyond the scope of what I can write here.

4. The conclusion itself may be accurate, but the premises, as stated, are not valid and so do not deductively lead to that conclusion.

5. Wikipedia defines inductive arguments as reasoning from the specific to the general. While such reasoning is inductive, many inductive arguments do not have that form (as discussed by the Internet Encyclopedia of Philosophy), and even some of the examples in the Wikipedia entry are inconsistent with this definition.

6. The definition of probable can be ill-defined. Technically, a standard of more likely than not (i.e., 51% likely) is a probable conclusion. Depending on the situation, such an argument may be considered strong enough or it may be considered weak. I’ve alluded to this in past posts, including in the context of trade-offs, science vs. lived experience, and our predictive algorithms.

7. Lawyers can tell you that the Supreme Court has ruled that invoking one’s 5th Amendment rights cannot be used as evidence of one’s guilt or innocence in a criminal trial.

8. Beyond our own assumptions and beliefs

9. This would, of course, be leveraging heuristics in the service of our confirmation bias.