How To Interpret Interpretivism?

Imagine trying to understand why a bird flies south for the winter. You might guess it is cold, or that it is hungry, or that it is following a map in its head. You are not measuring the bird's temperature or counting its heartbeats; you are guessing its thoughts and goals to make sense of its actions. This way of thinking, known as the "intentional stance," is how humans naturally interact with the world. We treat things—from pets to computers—as if they have beliefs and desires, because doing so helps us predict what they will do next. For decades, the philosopher Daniel Dennett has championed this approach, arguing that we can understand complex systems by treating them as rational agents. However, a long-standing puzzle has lingered in his work: exactly what kind of behavior counts when we make these guesses? Does it matter only what we can see on the outside, like a hand waving or a foot stepping? Or does it matter what is happening on the inside, like a racing heart or a chemical change in the brain?

A new paper by Gabriel Malagutti, a researcher at the University of Lisbon, tackles this confusion head-on. Malagutti argues that the debate over whether to look only at the surface or to dig into the machinery is not a contradiction in Dennett's thinking, but rather a sign that he is using two different tools for two different jobs. The paper suggests that when Dennett talks about what an observer can reasonably guess based on what they can see, he is making a point about knowledge and evidence. But when he talks about what a system actually is, deep down, he is making a point about reality itself. By separating these two ideas, the author claims we can finally make sense of Dennett's seemingly contradictory statements and apply his ideas more clearly to modern challenges, such as understanding artificial intelligence.

The core of the confusion lies in how we define "behavior." Some thinkers argue that only the visible, external actions of a system matter. If a person raises their hand, that is behavior. If their heart rate speeds up inside their chest, that is hidden and irrelevant to understanding their intent. This view is called peripheral behaviorism. Others argue that everything is behavior, including the internal workings of the body and the brain. If a machine has a secret internal list of instructions that drives its actions, that internal list is just as much a part of its behavior as the buttons it presses. This is called molecular behaviorism. Dennett has been accused of switching back and forth between these two views, sometimes saying only the outside matters, and other times insisting that the inside counts just as much. Critics have suggested that Dennett simply changed his mind over the years, but Malagutti proposes a different solution: Dennett was never inconsistent; he was just speaking about two different levels of understanding.

Malagutti suggests that the "outside-only" view is actually an epistemic thesis, which is a fancy way of saying it is about what we can know. If you are standing across the street watching a person, you can only see their movements. Based on those movements, you are justified in guessing they are happy or angry. Your guess is based on the evidence available to you. If you later get access to a lie detector or an MRI scan that shows their internal state, your guess might change. The paper argues that this shift doesn't mean the person's reality changed; it means your access to the evidence changed. Therefore, the "outside-only" approach is not a rule about what the world is made of, but a rule about what an observer is allowed to conclude given their limited view. It is a practical guide for how to interpret a system when you cannot see inside it.

In contrast, the "inside-and-out" view is presented as a metaphysical thesis, which is a claim about what actually exists in the world. Malagutti argues that when Dennett talks about the true nature of a system, he believes that the internal processes are real and that they determine what the system actually is. If a computer is running a program that mimics a human, but the internal code is just a giant list of pre-written answers with no real understanding, then it does not truly have beliefs, even if it looks perfect from the outside. The internal machinery matters for the truth of the matter. By separating the "what we can know" from "what is real," the paper resolves the tension in Dennett's work. It explains why he sometimes insists that passing a test of conversation is enough to treat something as a thinker (because that is what we can know from the outside), while at other times he insists that the internal structure must be right for it to truly be a thinker (because that is what is real).

This distinction helps clear up famous philosophical puzzles that have troubled Dennett's critics for years. One such puzzle involves a hypothetical machine called a "Blockhead," which is a giant lookup table that can answer any question perfectly but has no real mind inside. Critics argued that if Dennett only cares about the outside, he would have to say this machine is a thinking person, which seems wrong. Malagutti explains that Dennett's response to this is to say that such a machine is physically impossible to build. But more importantly, the paper argues that Dennett is addressing two different questions. If you are asking, "Can I treat this machine as a person for the sake of conversation?" the answer is yes, based on its external behavior. If you are asking, "Does this machine actually have a mind?" the answer is no, because its internal structure is just a list, not a mind. The paper shows that Dennett is not confused; he is simply answering different questions with different tools.

The same logic applies to the famous Turing Test, a method where a human judge tries to tell if they are talking to a machine or a person. If the machine passes the test, it acts exactly like a human. Some have argued that Dennett believes passing this test proves a machine has a mind. Malagutti clarifies that Dennett sees the test as a way to justify treating a machine as a thinker, not as a magical proof that it is one. The test tells us that we have good reason to use the intentional stance with the machine. It does not, however, guarantee that the machine has the same internal reality as a human. This nuance is crucial for understanding how we should interact with artificial intelligence today. We can use the intentional stance to predict how an AI will act, even if we are not sure what is happening inside its code.

By redefining these concepts, the paper offers a unified way to read Dennett's decades of work. It suggests that the ambiguity in his writing is not a flaw, but a feature of a sophisticated approach that distinguishes between the limits of human observation and the nature of reality itself. The author concludes that this dual perspective allows us to be practical in our daily interactions with complex systems, while remaining rigorous when we ask deep questions about what those systems truly are. This approach does not require us to choose between looking at the surface or digging into the depths; it allows us to do both, depending on whether we are trying to understand what we can know or what actually exists.