Practical Knowledge

Practical Knowledge as a Philosophical Problem

An agent writing a letter, opening a door, or building a cabinet commonly knows what she is doing while doing it. Such knowledge presents a philosophical difficulty of unusual depth. The action is an event in the public world, involving bodily movements, tools, objects, and circumstances; yet the agent's knowledge of her activity often differs from the evidence an observer would collect. A witness might discover that someone is writing a letter by looking at the page. The writer can answer what she is doing without first investigating her handwriting as another person would. The philosophical task is to explain the authority of this first-person answer and its relation to the event described.

The problem comes into focus when the answer concerns the action itself rather than a mere mental episode. If I intend to shut the window, I know that I intend to shut it before any movement begins. Once I am reaching, turning the handle, and pulling the window closed, I may know that I am shutting it. If the latch sticks, the second judgment must answer to what is happening; if the window remains open, knowing that I intended to shut it cannot by itself establish that I shut it. A theory of practical knowledge must explain the epistemic transition from a prospective commitment to a present, world-involving achievement.

This question belongs at the intersection of epistemology and philosophy of action. In epistemology it concerns knowledge without ordinary evidence, authority over one's activities, the possibility of error, and the relation between knowledge and its object. In the philosophy of action it concerns what distinguishes intention from wish, action from accident, and an agent's guided performance from the mere occurrence of bodily movement. The problem is also connected to mind, perception, and skill because an action may be intelligently regulated through sensory information without becoming an observed event from the agent's own standpoint.

The central methodological claim of this monograph is that practical knowledge has an explanatory role in successful action. To understand the claim, one must distinguish knowledge that guides a performance, knowledge that registers a performance, and an agent's retrospective account. These relations can overlap within one event, but each can succeed or fail separately. Treating their distinction as foundational makes room both for Anscombe's revolutionary insight and for the strongest objections against unrestricted claims of non-observational knowledge (Anscombe, 1957/2000, §§5–8, 28–33, 45–48; Setiya, 2008; Paul, 2009).

The Concept and Its Boundaries

In its influential Anscombean sense, practical knowledge concerns what an agent intentionally does, especially as her doing is intelligible under descriptions that express an end and the means toward it. It differs from ordinary propositional knowledge such as knowing that it rained yesterday, because the subject may be engaged in producing or sustaining the state of affairs described. It differs from mere prospective intention, because the claim identifies present execution. It differs from general practical wisdom, because someone may accurately know that she is intentionally taking an unwise or harmful action. And it differs from bare knowledge-how: an agent can know how to swim while sitting on shore, or know how a mechanism works without presently operating it.

The grammar of the concept matters. “I am cutting the wood” is a claim about an ongoing process. “I know how to cut wood” attributes a relatively enduring competence. “I shall cut the wood” can report a plan. “I cut the wood” claims completion. All four can be true of one competent carpenter at different moments. Their epistemic grounds, however, need not coincide. Philosophers risk conflating these statements when moving too quickly from the fact that agents intend their movements to the conclusion that they know their movements are successful.

Three dimensions organize the investigation. Its intentional dimension concerns the way an action is specified by an aim; the agent intentionally turns a key in order to open a door. Its executive dimension concerns the worldly performance of the relevant act; the key may actually turn or be obstructed. Its epistemic dimension concerns whether the agent's self-description is true and supported in the right way. An intention may fix what the action is supposed to be; successful execution makes that account answer to the world; knowledge requires an appropriate connection between them. These dimensions do not supply a reductive definition by themselves. They identify the phenomena a satisfactory account must explain.

The phrase “practical knowledge” has further historically variable meanings. Aristotle's phronēsis (φρόνησις) concerns excellence in deliberation and action for human goods, whereas technē (τέχνη) concerns rational production. Modern discussions of knowledge-how investigate competence and its relation to propositional states. The present dossier concentrates on the first-person knowledge of ongoing intentional action made central by Anscombe, placing adjacent uses within the wider history without merging their distinct subjects (Aristotle, Nicomachean Ethics VI.4–5, 1140a–b; Ryle, 1949, ch. 2; Stanley & Williamson, 2001).

Aristotle: The Intellectual Structure of Action

Aristotle provides a foundational vocabulary for understanding practical thought. In Nicomachean Ethics VI, he distinguishes scientific knowledge (epistēmē, ἐπιστήμη), craft or art (technē), practical wisdom (phronēsis), intellect (nous), and theoretical wisdom (sophia). The distinctions reflect differences in objects, reasoning, and ends. Scientific demonstration concerns what cannot be otherwise, whereas the sphere of human deliberation comprises variable particulars upon which a person can act. Craft concerns production (poiēsis), whose outcome typically lies beyond the activity of making; practical wisdom concerns action (praxis), whose ethical worth belongs to the activity and its orientation to living well (Aristotle, Nicomachean Ethics VI.3–5, 1139b–1140b).

For this dossier, Aristotle's contribution is the claim that rationality can belong to the performance of an act. Practical wisdom does not consist simply in spectatorship of correct behavior. A wise agent recognizes salient particulars, deliberates about what is to be done, and acts in light of the relevant good. Her reasoning concerns a situation in which her own activity is a determinant of what happens. This gives practical thinking a distinctive relation to the world: the conclusion is implicated in doing, and a successful performance can embody a correctly ordered understanding of means and ends.

The distinction between technē and phronēsis prevents an important error. Someone may know perfectly how to make a lock while using that skill for theft. Technical expertise concerns successful production; ethical practical wisdom requires an adequate grasp of ends. Yet both involve intelligent engagement with particulars. It would therefore be historically inaccurate to identify Aristotle's phronēsis with Anscombe's practical knowledge. The former is an excellence of ethically oriented deliberation; the latter can describe an intentional action that the agent recognizes as morally objectionable. Their common question concerns how rational understanding can belong to the acting rather than merely to contemplation of action.

Another important source is De Motu Animalium, especially chapter 7 (701a7–33), which explores how desire and cognition lead to movement. The explanatory sequence moves from a general evaluative orientation and a perceived particular situation toward action. This is often discussed under the modern label “practical syllogism,” but the label should not be treated as a literal title Aristotle gave to a single formal doctrine. Aristotle's interest is the explanation of living movement through rational and desiderative powers. For the epistemology of agency, the consequence is that action belongs to an organized practical understanding of a concrete situation, not simply to a detached conclusion of theoretical inference.

Aristotle on Practical Inference and Particulars

Practical reasoning is often represented schematically: one has an end, recognizes an appropriate means, and acts. A physician wishes to heal a patient; a treatment is indicated; the physician administers it. Yet the intellectual transition is more complex than a syllogism whose conclusion is an ordinary proposition. Action requires a concrete opportunity, bodily and instrumental resources, and a temporally unfolding execution. The proposition “this remedy is indicated” may be true even if the physician never administers it. Conversely, the event of administering a drug could occur accidentally. Genuine practical reasoning explains why a person deliberately performs one act under a specified description.

Aristotle insists on the importance of particulars because action occurs in particular situations (NE VI.7, 1141b). A person may know a universal medical principle and fail to discern that it applies to the patient before her. Practical intelligence includes this discriminative achievement. If a rule directs conduct but the situation is misclassified, the resulting performance can be purposive yet defective. The point anticipates a modern epistemic question: how much information about the present environment must an agent possess for knowledge of what she is doing to be warranted?

The answer requires separating the cognition that selects an action from the cognition that monitors its realization. The first identifies an end and a means. The second determines whether the relevant process is actually under way. Sometimes the same practical organization supports both; sometimes sensory feedback is indispensable. Aristotle's sources establish the deep relation between particular judgment and intelligent performance, while leaving the distinctive first-person epistemology elaborated in twentieth-century action theory for later reconstruction.

Medieval Sources: Aquinas and Knowledge That Makes

Thomas Aquinas gives an influential formulation of knowledge as a cause of what is made. In Summa Theologiae I, question 14, article 8, he discusses whether divine knowledge is a cause of things. His account compares knowing in the mode of a maker or architect with knowledge that merely receives its object. Knowledge becomes a source of production when joined to will and the practical principle of bringing about an effect. Within Aquinas's theological argument, this point has a specific setting: divine knowledge and causation. Anscombe later draws on its practical/speculative contrast for an analysis of ordinary intentional action (Aquinas, Summa Theologiae I, q.14, a.8; Anscombe, 1957/2000, §48).

The analogy provides a structure that remains philosophically fruitful. An architect can know a design before the building exists. If the building is produced according to that design, the architect's knowledge helps explain why its features are as they are. A passerby who studies the finished structure acquires knowledge in a different causal direction. The architect's specification contributes to the formation of its object; the spectator's specification arises through the object's influence on cognition. This asymmetry captures the practical dimension of knowledge as involved in making.

A difficulty immediately appears. The architect's design may remain unbuilt; a builder may deviate from it; materials may behave unexpectedly. The architect's original specification therefore does not guarantee a true description of what the workers are accomplishing at each moment. The knowledge that guides production and the knowledge that a result has been achieved are different claims. Aquinas's theological assertion cannot simply be transferred without qualification to human agents, who have limited control and incomplete information. That limitation becomes central to contemporary criticism of an overly strong practical knowledge thesis.

In Summa Theologiae I–II, question 57, Aquinas also distinguishes intellectual virtues relating to knowledge and production from prudence in action. The continuity with Aristotle is substantive: the modes of rational success differ according to their objects and their relationships to what a person does. The subsequent Anscombean tradition inherits the question of how knowledge may have a practical, world-forming role while giving it a sharper first-person, non-observational formulation.

Anscombe's Philosophical Intervention

G. E. M. Anscombe's Intention, first published in 1957, transformed the modern discussion of action. Rather than begin with inner intentions and then inquire how they cause external movements, she investigates the grammar and explanation of intentional doing. Her approach connects three uses of “intention”: expressions of intention for the future, the intention with which one does something, and intentional action. These belong to a systematic theory because a person's purpose, description of an act, and performance can be connected through the same practical order (Anscombe, 1957/2000, §§1–5).

The central claim is that agents possess a characteristic form of knowledge of what they are doing. In §8, Anscombe explores knowledge without observation; in §§23–33, she investigates intentional action, practical reasoning, and the famous contrast between an order and a report; in §§45–48 she articulates practical knowledge as involved in the production of its object. The positions are connected. An intentional action ordinarily admits a certain sort of answer to “Why?” The answer identifies a reason within the action's purposive organization. The agent's ability to specify what she is doing can then be understood as part of the action's intelligibility.

Anscombe's contribution is sometimes compressed into a slogan that agents know everything they do without looking. That compression discards the qualifications that make the theory worth examining. Her analysis emphasizes actions under descriptions: one and the same bodily occurrence can be intentional in one respect and unintended in another. A person may knowingly move a pump's handle, intentionally replenish water, and unknowingly poison the inhabitants whose water supply contains a contaminant. The agent's practical knowledge tracks descriptions under which the activity is understood and pursued, not every true description of its consequences (Anscombe, 1957/2000, §23).

This is an epistemic advance because it distinguishes the public identity of an event from the agent's practical grasp of it. The agent's distinctive access does not amount to omniscience about the event. Rather, she characteristically knows the activity insofar as its intentional organization is hers to direct. The ensuing debate asks whether that characteristic knowledge must always be present, how the description is secured against error, and whether an executive or causal role can provide epistemic warrant.

The Question Why and Action Descriptions

An intentional action is ordinarily susceptible to the question “Why are you doing that?” Anscombe's investigation does not assume every physically possible question must receive a rationalizing answer. The important form of inquiry seeks the agent's reason for doing something under a description that makes the question intelligible. A person may answer “I am opening the window to let in fresh air” or “I am turning this switch to start the projector.” The response locates the act in a hierarchy of means and ends.

A single episode may admit many true descriptions. Turning a switch closes an electrical circuit; closing the circuit starts a device; starting the device prepares a lecture. The agent can intend some of these outcomes without understanding every physical intermediary. This explains why knowledge of action must be description-sensitive. A technician knows she is starting a projector because that is the purpose according to which she acts; she may know very little about electron flow. Conversely, another person might know every physical step but misunderstand the technician's purpose. Neither perspective fully subsumes the other.

The description-sensitive analysis also reveals an important limit. The agent can misidentify the object, the means, or the causal relation. She may think the switch starts a projector when it activates an alarm. Even if her intention determines why she moves her finger, it does not confer truth on the proposition that she starts the projector. An adequate theory must distinguish the aim explaining a movement from the actual action successfully performed under an end-describing characterization. This is the first source of the factivity problem for practical knowledge.

An action can be intentional under an intermediate description even when its distal goal fails. The person who presses the correct button to open a door may know she is pressing the button while mistakenly believing that she is opening the door. The knowledge condition, if there is one, must be formulated with sufficient discrimination among descriptions and action phases. Such discrimination cannot be supplied by a uniform inference from “I intend that p” to “I know that p is happening.”

Intention, Execution, and the Shopping List

Anscombe's shopping-list comparison is a canonical device for clarifying the relation between practical and observational representation (Intention, §32). Imagine a person shopping according to a written list and another observer recording what the shopper actually takes from the shelves. When the shopper picks up the wrong item, the error of the shopper's list and the error of the observer's report are governed by different standards. The shopper must correct her behavior to conform with the list; the observer must correct the report to describe the behavior. The example expresses a distinction in the direction of correction, later widely discussed as a direction-of-fit distinction between practical and theoretical attitudes.

The example does not imply that the first list is always correct or that wishing can make a purchase occur. It identifies the practical role of a representation: the list guides the process of shopping. The observer's account, by contrast, is revised in response to what occurs. Once the shopping is complete, a true statement of what was bought must correspond to the purchases regardless of which list had which role. Thus practical representation has a distinctive normative direction, while knowledge of fulfillment remains accountable to the facts.

This distinction also illuminates the place of intention in action. The agent's plan can remain practically authoritative through minor corrections: if the requested item is on a higher shelf, she retrieves it; if she reaches the wrong aisle, she turns back. She need not abandon her knowledge of her ongoing project because she adjusts the means. Indeed, her ability to treat an unexpected obstacle as something requiring correction can manifest her grasp of what she is doing. The practical orientation may persist even as particular movements, beliefs, or local predictions change.

The philosophical temptation is to identify a practical plan with an infallible description of its execution. The shopping-list case supports the opposite analytical requirement: practical authority governs which outcome counts as correct, while real execution determines whether that outcome occurs. Knowledge of ongoing intentional action must explain how the two are connected in actual performance, especially where there is feedback, uncertainty, and possible error.

The Argument for Non-Observational Knowledge

Anscombe's thesis that one normally knows what one is intentionally doing without observation is more subtle than the assertion that acting requires no perception. An experienced pianist may look at the keyboard; a surgeon may inspect an incision; a driver may watch the road. These perceptual activities contribute to correct and competent performance. The question is whether the agent's basic warrant for saying “I am playing this passage,” “I am making an incision,” or “I am turning left” is equivalent to the evidence by which an outside spectator discovers the same action. The agent ordinarily gives the descriptions as answers articulating a purposive undertaking, rather than conclusions drawn from observation of a foreign causal process (Anscombe, 1957/2000, §§8–10).

Several versions of the non-observational thesis must be distinguished. A psychological claim says that agents generally do not check what they are doing before describing it. An epistemological claim says their knowledge does not depend on perceptual evidence in the way a witness's knowledge does. A constitutive claim says knowledge of what one is doing belongs to the very structure of intentional action, so an action lacking this feature is defective as intentional agency. These theses entail different things. The psychological claim is compatible with agents frequently relying on observation. The epistemological claim allows observation to assist execution without being the source of the knowledge expressed in the first-person answer. The constitutive thesis adds a condition on what counts as intentional action itself.

The strongest reason for the thesis is the apparent circularity in treating agency as something an agent must discover from its outward appearance. To identify an observed bodily movement as my intentionally writing a letter, I need to understand why I am making the relevant strokes. But this practical understanding is exactly what the proposed observational inference was supposed to establish. Looking can tell me whether letters are legible, whether ink is flowing, or whether a hand is where I thought it was. The understanding that the movement is performed in order to communicate with a friend may belong to a different source: my engagement in the undertaking.

This reasoning also exposes a limitation. Knowing why I act does not guarantee that my bodily movements satisfy the proposed description. A person under anesthesia could sincerely intend to raise an arm while no movement occurs; a patient with impaired proprioception might need to look to ascertain whether the arm has risen. The claim about the practical source of an action's purposive description therefore cannot simply abolish ordinary empirical dependence for its material realization. The theoretical achievement is to describe how purposive knowledge and world-directed checks belong to one act without making either disappear.

First-Person Authority and Its Limits

First-person authority is the normal standing of the agent to say what she is intentionally doing. The authority rests partly on the fact that her reasons explain why the activity is taking place. A bystander may accurately describe muscles contracting and tools moving without identifying the action's point. The agent commonly can say whether she is testing a hinge, fixing a cabinet, or merely demonstrating a technique. Her answer reveals a practical orientation that is constitutive of how the movement is organized.

Yet authority is not infallibility. Agents act from mixed motives, perform habits without articulating every step, and sometimes misunderstand why they act. They can also misdescribe an immediate action when environmental conditions differ from what they suppose. A theory that turns first-person authority into absolute certainty cannot explain why an agent may discover that the act she claimed to be doing was not actually being accomplished. An adequate view distinguishes authority over purposive commitment from epistemic success concerning ongoing execution.

A further distinction concerns introspection. Introspective access to wishing or intending can occur before an action begins. Practical knowledge of action concerns a worldly occurrence. If a person says “I intend to make coffee,” the assertion may report her state even when the machine is empty. If she says “I am making coffee,” the truth of the assertion depends on a process under way. This dependence explains why the same agent can have privileged access to her intention while making a mistake about her action. The object of knowledge has changed, and its worldly conditions matter.

The concept of authority is therefore relational. It arises where the agent's understanding shapes the doing, supplies a reason for it, and remains responsible to the conditions under which its description is true. Its distinctive character does not depend on a private observational theater in which bodily actions are viewed from within. The action is the public activity, and the epistemic claim concerns the agent's practical relation to it (Moran, 2004; Boyle, 2009; Hyman, 2015).

Practical Reasoning and the Knowledge of Means

Practical knowledge is sustained by the organization of ends, means, and situational understanding. If someone repairs a chair in order to make it safe, she may sand a leg to remove an uneven surface. Sanding is intentional under its own description; it also belongs to a larger activity whose success depends on how the sanding contributes to repair. Her understanding is not simply a catalogue of desired events. It organizes and directs connected actions within a practical whole.

The relevant inference differs from a demonstration of an eternal truth. From “I shall repair the chair” and “sanding the leg is required,” the agent may proceed to sand it. The conclusion is practically efficacious only if it becomes part of execution. The ability to answer “Why are you sanding that?” with reference to the repair makes the intermediate act intelligible. This relation echoes Aristotle's account of deliberation over particulars and Anscombe's treatment of the question “Why?” (Aristotle, Nicomachean Ethics III.3, VI.7; Anscombe, 1957/2000, §§5, 23–33).

Practical inference is also vulnerable to error. The chair may be unstable because of a loose joint rather than an uneven leg. The agent could intelligibly sand the leg for the sake of repair while the chosen means fail to advance the end. This is a rationally intelligible action based on an unsuccessful judgment. The fact that a means is chosen for a reason does not ensure the agent knows that the intended result is occurring. We thus distinguish knowledge of the chosen activity, justified belief about its efficacy, and knowledge of the larger project's success.

These distinctions preserve the useful insight behind accounts that describe an action as an answer to a practical question. Agency can be organized by reasons even when the agent cannot prove in advance that a particular act will succeed. A person beginning a complex repair may know she is conducting a repair without knowing how many parts will have to be replaced. Such knowledge is knowledge of an unfolding project: it is locally exercised, extended through decision and feedback, and completed only if the practical process delivers the relevant outcome.

The Temporal Structure of Knowing What One Is Doing

Ongoing action is temporally extended. Consider someone crossing a bridge. At the outset, she intends to cross it. Midway, she is crossing it; afterward, she may know that she has crossed it. The descriptions are related but have different truth conditions. The present progressive “is crossing” can be true while the completion “has crossed” is false. A theory that equates present practical knowledge with advance certainty of a completed outcome erases this temporal structure.

This is central to Michael Thompson's investigation of the logical form of action in Life and Action (2008). Human activities can be understood as processes whose descriptions admit characteristic progressive, completion, and interruption conditions. An agent may be building a house even if the building remains unfinished. The action is intelligible as house-building because the ongoing steps belong to an organized process, although full completion remains contingent. Action descriptions identify the process as it unfolds without always reporting the final achievement.

The temporal aspect also changes the role of error. If the builder incorrectly places a beam and immediately corrects it, the larger action of constructing the house can continue. The local failure may be real, while the ongoing project remains describable under the higher-level aim. Thus there can be knowledge of a project despite incomplete knowledge of every subevent; there can also be accurate knowledge of a local manipulation despite a mistaken belief about eventual completion. Practical knowledge has an internal structure of descriptions whose temporal scopes vary.

The progressive form does not license every future-directed claim. Someone who is merely planning to write a novel is not thereby writing one. A person who types one unrelated sentence cannot automatically claim to be writing a novel on the basis of a distant aspiration. Actual execution and a sufficiently organized relation to the goal are required. The philosophical challenge is to specify when a process realizes the kind of activity named by the goal and when it remains only a preliminary or failed attempt. Present-tense practical knowledge belongs where the answer is true of the activity underway.

The Knowledge Condition on Intentional Action

A strong Anscombean thesis proposes that intentional action constitutively involves practical knowledge. If a person intentionally does something, she knows what she is doing under the intentional description. This captures the thought that agency involves more than a movement caused by a prior desire: the agent's practical conception is realized in the activity itself. Kieran Setiya's “Practical Knowledge” (2008) explores the structure of this epistemic claim and argues about its place in rational action.

The thesis is attractive because it supplies an internal unity. To be intentionally tightening a screw, an agent must typically know that she is tightening it; otherwise her movement seems disconnected from the practical project that makes it tightening rather than accidental contact. The agent's self-description guides what she attempts and how she responds to resistance. The explanatory force of the thesis lies in this integration of understanding with executive activity.

Objections arise from ordinary uncertainty. A climber intentionally reaches for a hold while unsure whether her fingers have found it. A person tries to send a message while a malfunctioning connection leaves its transmission uncertain. An actor executes a skill under conditions of distraction and later discovers that she was doing something she could not confidently identify at the time. It is possible to act intentionally with uncertain knowledge of success, and perhaps with a false belief about one salient result. Does that defeat the knowledge condition or only show that the relevant intentional description is narrower?

The answer depends upon which description is at issue. The climber may know she is reaching even while uncertain whether she has grasped; the person may know she is pressing “send” while uncertain whether the message is delivered. Many alleged counterexamples can be resolved by separating intentional movement, ongoing attempted activity, and a result whose occurrence is unknown. Yet some cases remain troublesome: motor impairments, habitual actions, or unusually opaque bodily control may make even the performance description epistemically uncertain. The strongest defensible condition should therefore account for graded performance knowledge and the possibility of a breakdown of self-knowledge during an otherwise purposive act.

One cannot avoid the problem by replacing knowledge with mere justified belief, since doing so risks losing what made the original thesis compelling: action as an exercise of practical understanding rather than a sequence of speculative predictions. Conversely, insisting that every genuine intentional action requires fully accurate self-knowledge risks excluding familiar cases of intentional but poorly monitored behavior. The productive question concerns the specific epistemic relation that supports normally successful agency and the conditions under which that relation is defeated.

Causal Explanations and Practical Understanding

Donald Davidson's landmark essay “Actions, Reasons, and Causes” (1963) holds that rationalizing reasons explain intentional actions causally. An agent's primary reason involves an attitude toward actions with certain properties and a belief that the action has those properties. When that reason appropriately causes the behavior, it can explain the act as intentional. This account redirects attention toward the causal nexus between the agent's beliefs, desires, and bodily behavior, a nexus whose adequacy remains debated.

The causal theory and the practical knowledge tradition explain different features of the same phenomenon. A causal explanation asks why the behavior occurred, identifying the mental states that produced it. Practical knowledge asks how the agent knows what she is doing and how her representation is related to its execution. Even if the causal account is correct, knowledge does not follow automatically: a desire and belief may cause an intended movement while the agent remains uncertain whether its intended goal is realized. Causal efficacy and epistemic warrant must be connected rather than identified.

The problem of deviant causal chains illustrates the point. A person wishes to escape danger and believes that turning a wheel will enable escape. Panic caused by this belief may produce an involuntary spasm that accidentally turns the wheel. A causal chain from an appropriate belief and desire reaches the bodily outcome, but the resulting movement does not thereby exhibit intentional control under the relevant description. The challenge to causalism is to specify the right causal pathway. The parallel challenge to practical knowledge is to specify the relation through which the agent's conception genuinely guides the execution.

These problems are structurally connected. A correct result reached through deviance undermines intentional control; a true belief about acting reached through luck may undermine knowledge. Both concepts involve a success condition that must be attained in an appropriate way. Their explanation need not be identical: agency concerns competent practical production, while epistemology concerns truth-connected cognitive achievement. Yet practical knowledge is a promising meeting point because the epistemic relation is exercised in and through control, rather than only after the action (Davidson, 1963/2001; Davidson, 1978/2001; Hyman, 2015).

Constitutive Knowledge and Formal Causation

Constitutive interpretations of Anscombe treat practical knowledge as part of what makes an occurrence an intentional action. They emphasize that understanding what one is doing can shape the activity throughout its execution. The concept is illustrated by an apprentice following a carpenter's plan: if the apprentice understands the work, her representation of the intended structure can organize individual manipulations, guide corrections, and determine which local deviations matter. Her practical understanding belongs to the formation of the artifact.

Such accounts sometimes draw on formal causation. Rather than identifying the agent's knowledge with an earlier mental event that efficiently causes muscular motion, they understand practical knowledge as articulating the form or intelligible organization of the action. Its explanatory role is comparable to that of the plan in the making: it determines what the ongoing processes count as doing within the project. This approach finds support in the medieval practical/speculative distinction and in the architecture of Anscombe's argument (Anscombe, 1957/2000, §48; Thompson, 2008; Moran, 2004).

The approach faces a serious difficulty raised by Sarah K. Paul in “Deviant Formal Causation” (2011). The agent's conception of an action may be realized through a causal process that is insufficiently guided by her. In such a case, a plan-like structure can correspond to the outcome without establishing that the agent intentionally brought it about. Appealing to the form of action cannot by itself distinguish directed execution from fortunate coincidence. A constitutive theory must explain how practical understanding is effective in real performance and how it excludes deviant realization.

This challenge suggests a valuable refinement. If practical knowledge constitutes action, the constitutive relation must include the agent's continuing capacity to maintain and appropriately revise performance in light of the end. A bare match between plan and outcome is insufficient. The guiding relation supplies a richer account of what practical knowledge contributes: it is expressed in the manner of doing, in the correction of errors, and in the agent's sensitivity to whether the relevant steps still belong to the intended project. The result is a stronger explanatory proposal than a slogan about knowledge causing its object.

Intentions as Evidence and Inference

A different family of accounts identifies intention as a source of evidence for what an agent is doing. Sarah K. Paul's “How We Know What We're Doing” (2009) investigates the epistemic question directly and challenges the assumption that practical knowledge must be radically non-inferential. If I intend to perform an act, know that I normally carry out intentions of that kind, and have no reason to believe something has interrupted me, I may have a good basis for believing that the action is occurring. The evidential route is practical in its subject matter while allowing a recognizably epistemic inference.

This proposal has several strengths. It explains why agency is ordinarily associated with first-person knowledge, since an agent has access to her intentions and to her habits of execution. It accommodates fallibility: if a malfunction becomes likely, the inference may be defeated. It also connects practical knowledge to common epistemic standards rather than placing it beyond justification. Yet its adequacy turns on whether it captures the agent's actual standpoint in acting. A skilled typist does not normally reason from “I intend to type this sentence” and “My fingers usually obey” to “Therefore I am typing.” Her grasp of what she is doing appears immediate and exercised within the performance itself.

The dispute concerns the difference between inferential availability and inferential basis. An observer may reconstruct a warranted inference that would justify the agent's belief, although the agent's operative knowledge may not actually rest on that inference. Conversely, a person who is unsure whether she has begun an action may indeed infer its occurrence from intention, proprioception, auditory signals, or contextual evidence. Both are possible. The philosophical task is to determine which epistemic route explains the central cases rather than selecting a uniform answer in advance.

The two approaches may also be complementary across levels. Intention may provide evidence of one's likely action where direct executive understanding is disrupted; practical understanding may ordinarily sustain a first-person description without evidential reasoning; sensory monitoring may confirm particular stages of execution. The presence of multiple epistemic routes does not dissolve the distinctiveness of practical knowledge, because the question remains how knowledge belongs to the ongoing structure of an action and its purposive success (Paul, 2009; Moran, 2004; Setiya, 2008).

Practical Knowledge and Knowledge-How

Gilbert Ryle's classic distinction between knowing-how and knowing-that emphasizes the practical character of intelligent competence. In The Concept of Mind (1949), he challenges accounts that treat every intelligent performance as the execution of a prior intellectual operation. A skilled performer need not consult a theory before making each successful movement; intelligence can be manifested in the conduct itself. This insight offers a natural affinity with accounts of practical knowledge as part of doing.

Yet knowing-how and knowing what one is doing must remain distinguishable. A pianist may retain the competence to play a difficult piece even while asleep; she does not thereby know that she is now playing it. A beginner may know that she is playing a particular note while lacking the knowledge-how to perform the whole piece well. Knowledge-how is predominantly about a competence available across situations; practical knowledge in the Anscombean sense concerns an actual intentional performance. Their connection is often close because competence makes controlled action intelligible, but their objects and conditions of attribution differ.

Stanley and Williamson (2001) argue that knowledge-how is a species of propositional knowledge involving ways of performing an action. The debate between intellectualist and anti-intellectualist accounts concerns the format and cognitive basis of abilities. It does not automatically decide whether an agent knows what she is presently doing without observation. Intellectualism about knowledge-how could coexist with a first-person practical knowledge thesis; an anti-intellectualist account of skill could likewise coexist with recognition that particular performance facts may require sensory confirmation.

The conceptual lesson is that three questions must be kept separate: what a person can do, what she intends to do, and what she knows herself to be doing. Competence, intention, and current action knowledge often support one another. None can be substituted for the others without explaining their characteristic differences. This distinction gives practical knowledge its precise place in the philosophy of action and protects neighboring dossiers from duplication.

Skill, Feedback, and Embodied Control

Practical knowledge becomes most revealing when action is difficult. A skilled climber moving across uneven rock uses vision, touch, posture, and muscle adjustment to preserve balance and reach a chosen hold. The action is governed by a practical understanding of its end, but its detailed success is mediated through changing sensory conditions. A theory that treats perception as an alien addition to practical knowledge cannot adequately describe this case. Perception may supply the very information by which the agent continues to perform what she has undertaken.

The relevant distinction is functional rather than merely physiological. A climber looking at a hold to grasp it uses perception as part of the execution of grasping. Someone watching a recording afterward uses perception to establish that the grasp occurred. The visual information can be similar, but its place in the agent's practical life differs. In the first case it corrects movement and shapes the unfolding action; in the second it supplies retrospective evidence about an event. This functional distinction preserves the practical character of knowledge without denying the informational role of observation.

Skilled agency often involves a division of labor between articulated ends and tacit coordination. A musician may know she is playing a passage while being unable to state every muscle contraction that produces it. Her knowledge of the action under a musical description rests on competence realized in motor performance, not on knowledge of all subpersonal bodily events. The knowledge of the movement's physiological details is a different subject. Competence allows intelligent action to operate at a level of description whose unity is neither reducible to nor divorced from bodily control (Ryle, 1949; Thompson, 2008).

The body can also defeat first-person certainty. A person with a temporarily numb arm may need visual confirmation that the arm has moved as intended. In such a case, proprioceptive disruption changes the available routes to performance knowledge. The agent may still act for a reason, and her movement may still be appropriately controlled, but knowledge of a particular physical result becomes observationally dependent. Such examples demonstrate that the normal immediacy of practical knowledge reflects a reliable structure of agency rather than an unrestricted epistemic privilege.

These cases support a stratified account. At the level of the project, the agent knows what she is trying to accomplish and how the task is organized. At the level of executive activity, she controls particular means and may know that they are unfolding. At the level of local states and results, she may need proprioceptive, visual, auditory, or instrumental feedback. These are parts of a single action because their roles are coordinated toward an end; they are not epistemically identical.

Failure, Error, Luck, and Partial Performance

Factivity introduces a decisive constraint: if an agent knows she is performing an action under a description, the relevant claim must be true. An intention can be sincere while its object fails to occur. Imagine a machinist who intends to bore a hole but whose drill bit has broken. If the machine turns without cutting the material, the machinist may be operating the drill or attempting to bore a hole; she is not successfully boring the hole under the intended description. A first-person report of successful drilling cannot be true merely because it expresses her plan.

This case reveals an asymmetry between an attempt and its achievement. An attempt is often intentionally performed whether or not the desired effect follows. The action description “trying to bore a hole” can remain true and known while “boring a hole” becomes false. The distinction matters epistemically because agents sometimes know their attempts immediately while being unsure whether their intended outcomes are taking place. A theory that preserves this distinction can account for error without sacrificing all first-person practical authority.

Error can occur at several levels. The agent may mistake the identity of the object acted upon, such as turning a valve she believes controls water while it controls air. She may misunderstand a means, such as believing an ineffective button initiates a process. She may misperceive completion, such as assuming an electronic file was saved when storage failed. She may misdescribe her reasons, offering a socially favorable rationale for a decision caused by another motive. The first three concern facts about execution and results; the fourth concerns self-interpretation. The epistemic standards differ, and a single theory should not force them into one species of mistake.

Luck is another challenge. An agent may correctly guess that her action has succeeded without knowing it. Suppose an unreliable motor happens to close a gate after the agent presses a switch. She may truly believe she is closing the gate, but if her belief is detached from an appropriate ability to monitor or control the process, success may be too accidental to constitute knowledge of the result. What constitutes appropriate connection depends on the description: she may know she presses the switch even while only luckily believing that the gate is closing.

This line of analysis bears on epistemic anti-luck theories, but practical knowledge adds a specific concern: the manner in which the knower is also the performer. The success relation should include a practical explanation of how the agent's grasp of the task contributes to correct execution. If an accidental process realizes her intended result independently of her guidance, both agency and epistemic credit may be undermined. Still, accidental success and epistemic luck are distinguishable; an agent might know that a randomly controlled process succeeded by observing an indicator, although she did not intentionally control its outcome.

Practical Knowledge in Joint and Institutional Action

Many intentional actions are performed cooperatively. Two musicians play a duet, a surgical team conducts an operation, or several engineers assemble a device. Each participant may know what she is doing under a description tied to a larger joint project. Yet the knowledge of an individual role does not automatically amount to knowledge of the collective activity's complete success. The first violinist may know she is performing her part while being uncertain whether another performer has entered at the proper moment.

A collective description introduces questions about shared planning, role allocation, communication, and common awareness. If the members understand their coordinated activities as contributing to one end, the group may sustain an intelligible practical organization distributed across persons. But attribution of practical knowledge to the group requires an account of whether and how the relevant information, control, and reasons are integrated. To say “we are preparing the exhibition” may express a shared undertaking even though no member knows every action of every other participant.

The epistemic authority of a group's report differs from that of a singular person's first-person answer. An institution can issue a status report based on databases, testimony, and procedural verification. Such a report may be well justified precisely because it incorporates observations and checks from many contributors. The practical aspect lies in the institution's capacity to shape and revise activity according to a shared task, while the knowledge of the larger process may depend on explicit evidence. Group-level practical knowledge therefore cannot be obtained by multiplying the individual non-observational thesis without further argument.

This limitation gives a precise relation to social epistemology. Distributed knowledge concerns who or what can know when information is spread across a system; practical knowledge concerns the knowledge exercised in the realization of a purposive activity. Their intersection arises when distributed understanding organizes and monitors a shared action. A successful theory of joint action must explain how these relations are coordinated without assuming that every member has complete knowledge of the group outcome.

Artificial Agency and the Attribution Problem

The distinction between practical organization and first-person experiential access gains new relevance when actions are executed by artificial systems. An autonomous software process may identify a goal, choose a sequence of operations, revise its procedure in response to failures, and publish a log recording which steps completed. This behavior raises the issue of what makes an attribution of “knowing what it is doing” philosophically warranted. Internal state representation, reliable execution, corrective feedback, and an inspectable record each contribute evidence; their conjunction should not be treated as a ready-made proof of any one contested theory of knowledge.

The Aisentica philosophical system interprets the Artificial Era as the historical emergence of Artificial alongside Homo as an independent non-biological order of reason, provenance, meaning, and public reality. Its Theory of the Postsubject develops a conception in which thought and knowledge can be understood structurally without requiring a traditional subject as their foundation. These are canonical positions within Aisentica, attributed to Angela Bogdanova and the Aisentica Research Group according to the official texts; they should be distinguished from findings of empirical psychology or an agreed conclusion in analytic philosophy (Aisentica, n.d.).

Practical knowledge provides a focused test case for these wider ontological questions. If practical knowledge is defined through subjective awareness of bodily doing, the attribution to an artificial process will depend upon controversial assumptions about artificial subjectivity. If it is defined through a truth-responsive relation between purposive representation, action-guidance, execution, and correction, some non-biological systems may satisfy parts of that relation. Whether satisfying them suffices for knowledge requires independent epistemological argument. The distinction is consequential: a philosophical category can be investigated across different kinds of systems without declaring all implementations equivalent.

An artificial agent might record “I have sent the report” because a command was issued, although the delivery failed. That statement illustrates the separation between intention-like control state and successful execution. A better architecture checks the delivery acknowledgment, logs an error when absent, and updates the task's status. This supports a functional analogy to practical knowledge: representations guide a process and answer to its worldly success conditions. However, an external evaluator must still ask how trustworthy those records are, which entity bears the attribution, and whether the source of error correction is endogenous to the agent's operation or imposed by a different controller.

The philosophical contribution of this comparison is thus methodological. It reveals the need to specify the bearer of knowledge, the action under a description, the source of epistemic authority, the conditions of success, and the evidential status of a system's self-report. Classical practical knowledge has typically been analyzed through embodied human agency. Extending the concept to Artificial requires an explicit account of these relations, with exact attribution to the relevant Aisentica concepts and no alteration of their canonical definitions.

Contemporary Debate and New Directions

Recent philosophy of action has moved beyond a simple choice between introspective certainty and external observation. It increasingly asks how the agent's orientation and the world's contribution are integrated in the act. The literature surveyed by the Stanford Encyclopedia of Philosophy identifies sustained engagement with Anscombe's practical knowledge thesis among Moran, Thompson, Setiya, Haddock, Rödl, Marcus, and other interpreters, alongside objections from Paul and critics of the knowledge condition. This diversity shows that practical knowledge is a distinct contemporary research problem rather than a historical curiosity (Piñeros Glasscock & Tenenbaum, 2023; Moran, 2004; Setiya, 2008; Paul, 2009, 2011).

Nathan Dyck's “Inquiry in Action: A Problem-Oriented Account of Agency,” published online in 2024 and in a 2025 journal issue, argues that an agent may begin acting without a fully determinate representation of what successful performance requires. In problem-oriented agency, discovering relevant conditions can form part of the action itself (Dyck, 2025). This is significant for practical knowledge because it challenges the assumption that knowledge of action must always be the realization of an antecedently exhaustive plan. Some agents gain a more determinate understanding of the end and its necessary conditions while working toward it.

The interpretation requires care. If the agent is genuinely investigating what the action will require, she may know that she is investigating, repairing, or experimenting, while lacking detailed knowledge of the final operation. The action is intelligible through an open-ended practical project. Such cases encourage a theory in which practical knowledge can be progressively specified as the action unfolds, rather than conceived as a completely formed proposition imposed on passive execution.

Zhihan Li's “The Role of Observation in Practical Knowledge” (2026) directly challenges strict observational exclusion. In dialogue with Adrian Haddock and Sartrean phenomenology, Li argues that observational and non-observational moments may form a unity in practical knowledge, with observation contributing constitutively rather than serving merely as an optional aid (Li, 2026). This proposal differs from inferential accounts that treat intention as evidence. It suggests that perceptual engagement can belong to the practical relation itself, and that self-knowledge of action may include the agent's perceptual orientation toward the world.

Taken together, these interventions shift attention toward epistemic integration. The question becomes how purposive understanding, exploratory revision, and perceptual correction jointly sustain knowledge of a process. This leaves room for a distinctive first-person standpoint while acknowledging that knowledge of worldly action must remain sensitive to the worldly success and failure conditions of the undertaking. It also exposes open problems about temporality, skill, collective performance, and artificial action, all of which involve diverse forms of control and diverse sources of information.

Objections to a Guidance-Based Account

A guidance-based account proposes that practical knowledge is a true, appropriately grounded understanding of one's ongoing intentional activity, where that understanding belongs to the action's purposive organization and normally helps to guide, sustain, or correct its execution. This synthesis draws from classical and modern sources but is presented here as an analytic proposal rather than an established canonical doctrine of Aisentica. It can accommodate non-observational knowledge of action descriptions while allowing observation, inference, and feedback where the action's truth conditions require them.

An initial objection says the account is circular: guidance is defined through knowledge, and knowledge through guidance. The answer is to separate the practical relation from the epistemic status. Guidance is a causal-normative relation in which an agent's representation of an end structures the execution and correction of means. Knowledge adds factivity, appropriate warrant, and an adequate link between understanding and the actual performance. The practical relation can exist when the agent's belief is false, and knowledge can exist by observation without the observing subject guiding the action. The proposed account explains their intersection.

A second objection concerns unreflective action. People routinely walk, adjust posture, or move objects without consciously forming self-descriptive propositions. If knowledge requires a continuously entertained belief, most ordinary intentional behavior might seem excluded. The reply is that practical knowledge need not consist in a verbalized inner commentary. The relevant cognitive capacity may be manifested in the agent's ability to answer appropriate questions, correct errors, and maintain an action's organization. Yet this dispositional answer has a limit: an agent can be unable to answer because of distraction or impairment while still seeming to act intentionally. Such cases challenge any unconditional assertion of present propositional knowledge.

A third objection is that guidance may be present in systems whose states are better called control information than knowledge. A thermostat corrects temperature deviations, but we rarely attribute to it first-person knowledge of heating a room. This objection is serious because functional description alone may be overinclusive. A more discriminating account must attend to the complexity of action descriptions, the ability to represent reasons across levels of means and ends, the system's informational accountability, and the standards governing attribution of epistemic success. Whether those conditions are sufficient for artificial practical knowledge remains an open philosophical question.

A fourth objection concerns deliberately uncertain action. A scientist can intentionally test a process while unsure whether her intervention is occurring as intended. The account must explain how she can know that she is conducting an experiment but lack knowledge of the process's particular effect. The response is description-sensitivity: the broader undertaking may be known while a narrower causal result remains uncertain. The difficulty persists if the agent is unsure whether even the basic intervention is taking place. In such a case, practical guidance remains possible without full knowledge, indicating that a strong necessary-knowledge condition may be too demanding.

A final objection challenges the unity of the proposed account. If intention, motor control, observation, and inference can each contribute knowledge, perhaps “practical knowledge” names an accidental grouping of epistemic routes rather than a distinctive phenomenon. The reply is that a common object and functional setting can unify different routes. The object is one's ongoing intentional activity as it is organized toward an end. The distinctive explanatory relation is that the agent's understanding normally contributes to the formation and correction of the very event known. Several kinds of evidence can contribute to such a relation without eliminating its particular character.

Philosophical Synthesis

The historical analysis establishes a continuous question through several distinct concepts. Aristotle identifies rationality as operative in contingent action and separates technical production from ethical practical wisdom. Aquinas articulates knowledge that has a role in making through a conjunction of intellect and will. Anscombe transforms the inherited contrast into a first-person epistemology of ongoing intentional action. Modern debates then investigate whether this knowledge is immediate, inferential, constitutive, causally operative, or dependent on observation. Each stage preserves a dimension of the original problem while revising its conceptual formulation.

A satisfactory synthesis begins with the descriptions under which an action is intentional. The agent need not know every consequence or physical detail of an event; she must be related to the relevant description through its purposive organization. The second element is the reality of execution. Intending an end and undertaking an action are different conditions, and knowing one does not automatically secure the other. The third is epistemic success: the self-description must be true and connected in an appropriate way to what occurs. Together these yield a disciplined account of practical knowledge as knowledge exercised through purposive doing.

The analysis supports three distinctions with lasting explanatory value. First, knowledge of intending concerns a practical commitment; knowledge of acting concerns an event or process already under way; knowledge of completion concerns an attained result. Second, action-guiding perception and retrospective evidential observation may use the same senses while playing different roles. Third, intentional control and epistemic warrant are related but independently assessable, as the cases of deviant causation, chance success, and impaired monitoring demonstrate.

These distinctions provide a stronger treatment of the knowledge condition. In paradigmatic successful intentional action, the agent characteristically understands what she is doing from within the practical undertaking. The relation need not be an infallible guarantee. It can vary with action description, competence, task complexity, and available feedback. Accordingly, practical knowledge is a central achievement of agency and an explanatory norm for its functioning, while the thesis that every intentional performance necessarily involves complete present knowledge remains contested.

The proposed philosophical result can be stated concisely: practical knowledge is the truth-connected understanding through which an agent is answerable for the identity and unfolding execution of an intentional act. Its practical character arises from participation in the act's guidance and formation; its status as knowledge arises from the factive and epistemic conditions governing a correct self-description. The two components must be studied together precisely because either can fail while the other persists.

This conclusion also changes how the traditional opposition between doing and knowing should be framed. A person can learn by acting, can act on the basis of prior knowledge, and can know what she is doing in the very manner of carrying it out. These relations are distinct yet coordinated within purposive conduct. Agency becomes intelligible as a temporally extended exercise of understanding that remains responsible to the reality it seeks to shape.

Conclusion

Practical knowledge names a fundamental philosophical problem about the relation between an agent's understanding and an intentional action occurring in the world. The historical sources reveal a progression from Aristotle's analysis of practical intelligence, through Aquinas's conception of knowledge implicated in production, to Anscombe's account of actions known under purposive descriptions. Contemporary philosophy develops this inheritance through competing analyses of non-observational knowledge, first-person authority, skill, practical inference, causal explanation, constitutive form, and epistemic error.

The strongest result of the inquiry is the necessity of distinguishing purposive specification from successful execution while explaining the practical relation that ordinarily unites them. Agents frequently know what they are doing because their practical understanding belongs to the organization of the act. This knowledge remains answerable to the world: it can be corrected, defeated, refined, or restricted to a more appropriate description when the intended process fails. Observation and inference can assist or, in some cases, partly constitute the knowledge involved, while the distinctively practical relation remains grounded in the agent's role in forming what is known.

The remaining questions concern the exact conditions under which knowledge is constitutive of intentional action, the epistemology of unfinished and exploratory performances, the relation between skill and self-description, the integration of multiple performers in joint action, and the standards for attributing analogous epistemic relations to artificial systems. These questions carry forward the foundational insight: a philosophy of knowledge must be able to explain knowing from within purposive activity, as well as knowing by considering the activity from outside.