Hume: Habit of Mind and the Expectation of Causes

Imagine a hypothetical scene. Three times you sent a request in the evening, and three times received an irritated reply. On the fourth evening, you have not yet opened the conversation but already expect the same thing. Perhaps evenings really are inconvenient. Perhaps the request’s content, urgency, or other circumstances matter. But the expectation has already arisen and looks almost like knowledge of the future.

Another hypothetical person chose the same way of working three times and obtained a good result. Now they call the approach reliable. Both people move from what has happened to what does not yet exist. Where does confidence in that transition come from? And what exactly was repeated: the visible event, its conditions, or our story about it?

What This Article Is About

David Hume asks a question that is hard to eliminate simply by urging trust in experience: on what grounds does the past become a basis for expectation? We will briefly consider the distinction between relations of ideas and matters of fact, the problem of moving towards the future, and the significance of habit. We will then develop contemporary examples of our own: recurring conversations, a successful working approach, the coincidence of two events, and the need to make decisions without guarantees. These examples are not behavioural studies and are not attributed to Hume. His analysis is a philosophical account of human expectation; it neither makes every forecast equivalent nor exempts us from checking. At the centre is a useful tension: past experience is indeed necessary, but its strength depends on the precise conclusion we draw from it and how honestly we name the unknown.

1. Two Different Kinds of Confidence

In Section IV of the Enquiry Concerning Human Understanding, Hume distinguishes relations of ideas from matters of fact. Denying a mathematical relation may involve a contradiction; the opposite outcome of a factual event is conceivable without such a contradiction. Expecting an event therefore does not have the same foundation as proving a relation between numbers. Hume, Enquiry, IV, §§20–21.

For our reading, placing two sentences side by side helps: “In this table, the sum of three numbers equals the stated result” and “Tomorrow the participants will do the same again.” The first can be checked by calculation if the numbers and operation are defined. The second requires information about events, conditions, and participants. Even an error-free calculation does not prove it.

Imagine planning a meeting. If its start is fixed and the journey usually takes half an hour, someone calculates when to leave. The arithmetic may be precise. Tomorrow’s journey time remains an expectation. We often combine these parts into one confident “I will get there in time,” forgetting their different status.

This distinction does not make the plan bad. It shows where a margin is needed and where adjustment is possible. A mistake in addition is remedied by recalculation; a changed journey by new information or a change in action. If we try to solve the second question by increasing confidence in the first, our check misses its object.

The same structure appears in an argument. Someone says, “Since they refused twice, they will refuse now.” The number of previous refusals may be established precisely. A conclusion about the current event still requires a transition. Perhaps the request is the same and conditions similar; perhaps something important has changed. The word “therefore” does not render these possibilities unnecessary.

Our first practical thought is simple: certainty about part of a line of reasoning should not silently extend to the entire line. We can be confident in the record of past events, less confident in their explanation, and less still in the next outcome. Naming these levels does not mean abandoning a decision. It means understanding what supports it.

2. Between Repetition and a Law

Hume asks why an observed connection should continue beyond what has been observed. Proving it through past experience already presupposes the very transition whose justification is sought. In Section V, habit is considered as a principle forming expectation after repetition. Hume, Enquiry, IV, §§29–32; V, §§35–38.

Our hypothetical evening request shows the distance between three formulations. “The last three evening replies were irritated” describes a sequence. “Evening causes irritation” adds an explanation. “It is impossible to reach agreement in the evening” turns the explanation into an almost unconditional limit on the future.

Each subsequent formulation grows broader. To choose a conversation time, an initial working hypothesis may suffice: evenings currently seem inconvenient. To assert causation, we must distinguish other conditions. To declare impossibility, stronger grounds still are needed. Practical caution is often justified before a final explanation is available.

Imagine the opposite repetition: work finished faster every time a procedure changed. We may reasonably retain the procedure. But we need not claim that it is equally effective for every kind of task. If previous cases were small, new large-scale work may require further checking. This is our thought experiment, not an established effect of a particular method.

It therefore helps to ask what scale of repetition we actually observed. One person, one kind of request, one week, one set of conditions is not “always.” A narrow regularity can be practically valuable precisely because its boundary is named. It helps us act under similar conditions without pretending to describe the entire world.

An objection naturally arises: if we keep limiting a conclusion, will its usefulness disappear? Not necessarily. We can say “Under these conditions, this approach has worked so far” and use it again while preserving room for correction. Action needs grounds, but does not always need a proven law. Reducing verbal universality sometimes makes experience more transferable: it becomes clear what to retain and what to check when the situation changes.

3. Habit Helps Us Live and Can Close a Question

Familiar expectation saves us from reconsidering every case anew. This is our account of its function in a hypothetical example, rather than a measured psychological effect. If you know how to open a familiar door, you do not conduct a philosophical inquiry before every movement. If a familiar working approach is suitable, you use it until a significant difference appears.

But habit may start answering a question before an event occurs. In the opening scene, the person prepares a defensive response in advance and sends the request in an accusatory tone. The fourth conversation then already differs from previous ones. If it is difficult again, we cannot automatically call it a pure repetition of the same condition.

It matters not to declare the person themselves the cause of every failure. Perhaps the other person really did respond sharply again, regardless of the approach. Our question is more modest: did the event remain the same once expectation entered our action? Sometimes our own preparation, tone, or failure to clarify becomes part of a new situation. Sometimes no substantial influence can be found.

We can imagine a positive expectation. Someone trusts a familiar approach so much that they overlook a new detail of the assignment. If the work is unsatisfactory, checking the change in the task is useful, rather than immediately declaring past experience false. The approach may have been appropriate before and insufficient now. The two propositions are compatible.

In our reading, a mature use of habit does not require destroying it. What matters is noticing when it starts hiding a difference. Which sign tells us conditions have changed? Which part of the approach should be retained? Which step needs a fresh check? The answer may be very small: a different requirement appeared, a participant changed, the scope increased.

This turns familiar experience from a prohibition on change into a working foothold. We do not start from zero, but do not give the past an unconditional right to decide for the future. Habit remains useful while its scope is visible and new information can enter the next attempt.

4. Cause, Coincidence, and a Shared Condition

Two events can occur together without the first being a sufficient explanation of the second. This is our analytical question, rather than a ready-made observational finding. Imagine a hypothetical room: people argue more often when the window is open. We might suspect the noise. We might notice that the window opens during the hottest hours, when conversations are already harder. We might discover that the most difficult tasks are discussed at precisely those times.

One sequence admits several explanations. For action, reducing the noise and seeing what changes may be enough. A causal claim requires more careful distinctions. A hypothetical experiment may yield no clear result: the subject, duration, and participants changed at the same time. An honest conclusion would then be limited.

In the Treatise of Human Nature, Hume offers rules for judging causes, including consideration of event order, similarity between cases, and differences in circumstances. His historical list is not equivalent to contemporary research methods. Hume, Treatise, I.III.XV. What matters to us is that analysing expectation is compatible with disciplined discrimination, rather than indifference to every ground.

Let us distinguish three questions: what happened earlier; what regularly occurs together; what changes an outcome with other important conditions held constant? They do not always receive the same answer. Order may exclude some explanations, but order alone does not establish all the others. Repetition may strengthen expectation without yet identifying a mechanism.

Avoiding words that describe more than was observed is especially useful here. “After this” may be accurate. “Because of this” needs additional grounds. “Only because of this” excludes other conditions and therefore needs more still. The force of grammar should not replace the force of evidence.

An ethical boundary also emerges. Another person must not be made an object of a convenient experiment without their participation. If we change a shared agreement, they should understand what they are agreeing to. Our examples suggest voluntarily clarifying one’s own action and communicative conditions, rather than secretly testing another person’s character.

A good causal question complicates the picture only as much as a decision requires. It need not turn a simple situation into a research laboratory, but should not declare the first noticed connection a final answer either.

5. What the Strength of Evidence Means

Distinguishing grounds avoids a false equivalence: without an absolute guarantee, every explanation supposedly has equal value. In our hypothetical example, a forecast supported by several comparable cases generally has different practical grounds from a sudden guess. But “several” alone does not settle the question: distinguishable cases and a clear object of comparison are needed.

Imagine two people assessing the reliability of an agreement. One cites three messages from one participant in one conversation. The other cites three commitments fulfilled over different periods. These are different kinds of information. The first can show consistency in words; the second adds observation of outcomes. For a new decision, it helps to know what kind of trust we want to justify.

Quantity also fails to correct a mistaken unit of counting. One account repeated on ten pages is not ten independent events. If ten people retell one source, the information may be variously expressed, but its origin remains shared. This is our methodological distinction; the article presents no measurements of a particular media environment.

The unit may be simpler in everyday situations. What counts as an instance of delay: a message, a day, or a promised action? Remembering one event many times does not turn it into a series. If several delays occurred under substantially different conditions, a general conclusion must name that difference.

But a demand for precision should not diminish a single serious event. Sometimes one refusal or direct breach is enough to change one’s participation. Such a decision does not require proving someone’s universal character. We can say, “This action is unacceptable under our agreement,” without moving to “They always act like this.”

The strength of evidence thus relates to the strength of a claim. Deciding not to repeat a certain situation sometimes needs less universality than accusing someone of an unchanging nature. A practical boundary and a philosophical law require different grounds.

Our reading proposes asking “For which exact claim is it sufficient?”, rather than merely “Do I have enough facts?” This small question keeps experience away from two extremes: complete dismissal and unjustified expansion.

Hume presents another important tension: repetition does not add a visible necessary connection to observed objects, but changes the transition in the mind expecting the next member of a sequence. Academic commentary considers this distinction between external observation and internal expectation. Qu and Radcliffe, “David Hume,” §2.2. This is philosophical analysis, rather than a modern experimental measurement of habit’s strength.

Our thought experiment makes the problem tangible. Imagine an entirely unfamiliar device with two lights. After the first turns on, the second follows. Even after repeated observation, what appears before you remains a sequence of light signals. Naming the mechanism requires additional information about the device, an appropriate check, or a proposed model. More observations strengthen the grounds for expecting the sequence; they do not make a hidden mechanism directly visible.

Discovering a mechanism would not eliminate every question either. Suppose you saw the connection and established how it works. Prediction now rests on more definite understanding. But it assumes that the device remains functional and conditions match those checked. We need not investigate everything anew after each activation. We can simply name the conditions on which trust rests and the signs that would require returning to an inquiry.

The distinction particularly matters when confidence becomes an accusation. A hypothetical person repeatedly encountered silence after a request and now believes they see the intention to reject them itself. Yet repeated silence and intention remain different objects. We can decide on the basis of silence without turning an unknown motive into an observed fact. This does not weaken the other person’s obligation; it simply preserves the accuracy of our own claim.

A question about causes therefore requires more than a question about the next expectation. Sometimes we really need a mechanism. Sometimes a reliable working sequence is enough. But we need to understand which task we are addressing, so that success in one does not become a final answer to another.

6. A Scepticism That Does Not Prohibit Action

An uncertain future does not remove the need to choose. Hume distinguishes the practical use of experience from the philosophical question about grounds for the transition; these positions appear alongside one another in Section IV. Hume, Enquiry, IV, §31. We do not turn this distinction into a demand to abandon inquiry.

In a hypothetical scene, someone must choose when to have a conversation. They do not know for certain what response they will receive. They can consider previous meetings, ask for a convenient time, or choose a format permitting a pause. Each decision has limits. The absence of a guarantee does not make them equivalent.

Our practical question concerns acceptable consequences of error. If a choice is easily revisable, trying a small action and observing may be reasonable. If an error creates a substantial commitment, gathering more information and explicitly discussing the terms is useful. This is a contemporary editorial position, rather than a ready-made Humean decision theory.

We can name its measure without invented numbers. Which circumstances could change the decision? What is already known? What must be assumed? How soon will the result become visible? Can the next step be stopped if the assumption is not confirmed? Sometimes these questions provide enough grounds even without a precise probability.

There is a risk of hiding a decision behind “unknown.” If someone says, “I cannot know, so I choose nothing,” that too has consequences: a deadline passes, an agreement goes unanswered. Postponement may be reasonable, but should be named as an action with conditions and costs, rather than neutral nonparticipation.

The opposite risk is to treat the need for a decision as proof of the chosen explanation. “I need to act” does not mean “My hypothesis is certainly true.” We can choose a next step while openly retaining the explanation’s working status. New information will then not be experienced as an attack on an already proclaimed truth.

In this reading, scepticism gains a useful limitation. It questions a transition and helps match confidence to grounds. It need not stop the door, journey, conversation, and every everyday action until philosophy completes its dispute about induction.

7. Objections That “Habit” Cannot Remove

The first objection is obvious: many successful expectations rest on much more than a repeated external sequence. We understand a tool’s workings, know its operating conditions, and can explain the sequence. Our contemporary response is that these differing grounds should be considered. The article does not propose replacing every investigation of mechanisms with observations of frequency.

But knowledge of how something works also needs caution when transferred. A hypothetical person may correctly understand a familiar appliance without knowing that an important detail changed in a new version. An explanatory model strengthens the grounds without making an unfamiliar case already checked. Hume’s question remains a stimulus to name the exact transition being made.

The second objection concerns causation itself. Does analysing expectation mean there are no real causes? Academic discussions of Hume distinguish the origin of the idea of necessary connection from disputes about what he claims concerning causal powers. “Hume abolished causation” is too crude a formula. “David Hume,” section on induction and necessary connection.

The third objection: habit does not explain why one expectation should be preferred to another. Our contemporary development indeed does not solve the entire epistemological problem. It offers limited ways of distinguishing grounds in a particular case. A history of past success, comparable conditions, and opportunities for checking have practical significance, but this article does not prove an ultimate norm for every forecast.

The fourth objection concerns relationships: if I keep leaving uncertainty open, will I become too accommodating? Not necessarily. We can retain uncertainty about motives while setting a clear behavioural boundary. “I do not know why you failed to warn me” is compatible with “I am not prepared to continue working under these terms.”

Finally, reflection itself should not empty experience of meaning. Past events matter and can change our choices. Limiting a conclusion does not mean crossing out memory. The difference lies in whether we are speaking of what happened, a working expectation, or an unconditional law. The same experience may confidently support the first, cautiously support the second, and insufficiently support the third.

Philosophical honesty here preserves two movements: acknowledging that experience takes us beyond what is present, and not hiding that this transition deserves attention of its own.

8. What This Means in Practice

If you wish to investigate an expectation, you can start with “After this, what usually happens is…” Choose a small episode allowing observation and change. You need not turn to the hardest relationships or most consequential decisions.

First, list specific cases you actually remember or can check. Then name what stayed the same and what differed. In the hypothetical evening example, the time might have stayed the same while the content of the request differed. Perhaps you discover that you remember only difficult conversations and omitted calm ones. Perhaps, conversely, the sequence proves consistent. Neither outcome should be proclaimed in advance.

The next question is: which conclusion do you actually need? Do you want a more convenient conversation time, to understand irritation’s cause, or to decide whether to continue an agreement? These are different tasks. For the first, asking the other person’s preference sometimes suffices. The second may remain open. The third also depends on your boundaries, rather than response frequency alone.

If a small attempt suits the situation, change one available element and name an observation in advance. Instead of an evening message, for instance, agree on a discussion time. Then see whether the conversation happened and assent became clearer. Do not call one successful outcome proof of a discovered cause if other conditions changed simultaneously.

It helps to anticipate an unsuccessful result too. Another difficult reply need not mean no change is possible. Perhaps the altered element was not decisive. But promising that the next attempt must work is also unwarranted. It is acceptable to stop, retain the observation, and choose another way to participate.

The other person’s consent remains important. Our exercise does not propose provoking people, testing them secretly, or extracting admission of a supposed motive. It aims to clarify one’s own hypothesis and a permissible action.

At the end, you can write three sentences: what was observed; what you expect; what you do not yet know. The note need not become a daily ritual. Sometimes one episode is enough to hear the difference between experience and its overly confident extension.

9. The Past as a Foothold With an Open Edge

Let us return to the two hypothetical people from the beginning. One expects irritation, the other success. Their expectations are not necessarily wrong. The problem arises when expectation acquires the status of a future that has already happened and stops noticing conditions.

If the first clarifies the conversation time, they may receive a new answer. If the second checks the changed task, they may retain a useful approach and adjust one detail. Sometimes nothing substantial changes. Even then, expectation will rest on a more definite sequence, rather than a familiar feeling alone.

Our reading does not promise freedom from habit. It proposes working with its boundary. The past suggests a direction, but a particular new event still needs to be encountered. That encounter may be small: reading the current condition, hearing the current reply, checking the current action.

The right to trust matters too. We need not begin each morning by suspecting all accumulated experience. Remembering that trust has an object is enough. I can trust a familiar approach in a familiar task and check its suitability for a substantially different one. I can trust a person and clarify a new agreement. These actions do not contradict one another.

Hume’s question returns us to the moment of transition. We saw what happened. Now we expect what will happen. Sometimes the transition is justified as far as a decision needs; sometimes it has become too broad. Noticing that difference is our contemporary task, rather than a final victory over the philosophical problem.

Past experience remains valuable precisely when it admits continuation and correction. It need become neither an infallible law nor a collection of meaningless accidents. Between those extremes lies practical honesty: naming grounds, limiting a claim, and meeting a new case ready to see what the previous sequence has not yet shown.

Sources

1. David Hume. An Enquiry Concerning Human Understanding: IV, §§20–21, 29–32; V, §§35–38. Project Gutenberg.
2. David Hume. A Treatise of Human Nature: I.III.XV, rules for judging causes. Project Gutenberg.
3. Hsueh Qu and Elizabeth S. Radcliffe. David Hume, §2.2, induction and necessary connection. Stanford Encyclopedia of Philosophy, current 2026 revision; authorship checked against Author and Citation Information.