A Philosopher’s Guide to Thinking, Reasoning, and Pretending to Be Logical
Executive Summary
After centuries of what can charitably be described as “productive disagreement” and, less charitably, as “professional navel-gazing conducted in well-appointed seminar rooms,” philosophers, psychologists, and cognitive scientists have reached a consensus that would impress even the most quarrelsome parliamentary body: thinking and reasoning are probably different, though nobody can quite agree where the boundary lies. Several tenured professors have built entire careers insisting that the distinction is either (a) obvious, (b) nonexistent, or (c) the most important insight since Descartes realized he existed (which, frankly, set the bar rather low).
The terms overlap in everyday use, like two drunk uncles at a wedding reception. Everyone knows they are related, but watching them interact is both confusing and mildly concerning. However, upon closer inspection (and with sufficient coffee), thinking emerges as the broader, more catholic category of mental activity, encompassing everything from daydreaming about tropical vacations to solving differential equations to wondering whether you left the stove on. Reasoning, by contrast, is the prissy, rule-following cousin who insists on drawing conclusions from premises and won’t let you get away with logical fallacies at the dinner table.
Holyoak and Morrison (2012), in what remains the most-cited definitional hand grenade in the field, define thinking as “the systematic transformation of mental representations of knowledge to characterize actual or possible states of the world, often in service of goals” (p. 3). Notice what this permit: daydreaming about being a rock star counts as thinking (transforming representations of possible states), as does planning your taxes. Meanwhile, they characterize reasoning more narrowly as the process of “drawing inferences (conclusions) from some initial information (premises)” (Holyoak & Morrison, 2012, p. 5). Deduction, induction, and abduction all qualify, though abduction, as every philosophy student discovers at 2 AM, sounds exciting but is really just “inference to the best explanation,” which is about as thrilling as watching paint dry that might be a certain color.
Historically, Aristotle and Kant treated reasoning as the mind’s normative function, the part that should work correctly, even if it often doesn’t. Hobbes (1651), in a moment of refreshing clarity that must have annoyed his more verbose contemporaries, declared flatly that “Reasoning is but reckoning” (Leviathan, Part I, Chapter 5, para. 2), a computational view that made him either a visionary or a killjoy, depending on whether you enjoy the romance of imprecise thinking. Frege and Kant famously held that logic is normative, prescribing how one ought to think (MacFarlane, 2004, p. 24). This stands in delightful tension with psychology’s discovery that real human thought often deviates from these norms in creative and predictably stupid ways, a phenomenon known in the trade as “being human” and studied under the slightly more dignified rubric of “heuristics and biases” (Kahneman, 2011, p. 12).
Across disciplines, the distinction takes on different flavors, like a chameleon at a paint store. Philosophy of mind treats thinking as a general feature of consciousness (including nonpropositional imagery and those mysterious unconscious processes that somehow decide what you’re thinking before you think it), while reasoning is treated as the special case of rational thought involving propositional attitudes and inference. Epistemology and logic study reasoning in formal terms (deductive validity, inductive strength, and the kind of symbol manipulation that makes undergraduates weep), while thinking encompasses justification, belief formation, and mental content more broadly, essentially everything the logicians were too busy to notice.
Cognitive science, never one to let philosophical niceties get in the way of a good experiment, studies both: mental rotation tasks (Shepard & Metzler, 1971) demonstrate that people literally think with images, rotating three-dimensional shapes in their heads at speeds that correlate with the angle of rotation, proving that thinking is not merely linguistic (Kosslyn, 1994, p. 45). Meanwhile, syllogism tasks and probability problems reveal that human reasoning is systematically flawed in ways that keep both psychologists and comedians employed (Tversky & Kahneman, 1974, pp. 1124-1125).
First-person phenomena, those moments when you say “I think” or “I reason,” raise delightful questions about introspection and self-knowledge. Some philosophers (following Evans, 1982) argue that our belief-ascriptions are “transparent”: to know what you believe, you look at the world, not inward (p. 225). This is either profound or a recipe for gaslighting yourself into believing you like kale. Saying “I think that P” ascribes a belief; saying “I reason that P” comments on your method, often implying a justification that may or may not exist (Nisbett & Wilson, 1977, p. 233; more on their devastating findings later).
The role of language remains contested, like a disputed territory between two academically inclined nations. Fodor’s (1975) language-of-thought hypothesis posits an inner “Mentalese” in which thinking occurs as symbol manipulation. This view makes thinking look suspiciously like programming, which either vindicates AI researchers or reduces human cognition to something a Commodore 64 could allegedly accomplish (p. 78). Yet mental imagery studies (Kosslyn, Thompson, & Ganis, 2006) show that thought can be nonlinguistic: tennis players visualize serves, architects imagine buildings, and philosophers imagine themselves actually finishing that book chapter (pp. 14-17). The ensuing debate has generated enough academic publications to wallpaper the Library of Congress.
Methodologically, introspection has taken a beating. Nisbett and Wilson (1977) delivered the philosophical equivalent of a cold shower by showing that people have “near certainty” about their current experiences yet are “almost completely unaware” of the actual causes of their judgments (p. 233). In one famous study, subjects couldn’t explain why they preferred one pair of pantyhose over another (hint: position effects), yet confidently fabricated reasons (Nisbett & Wilson, 1977, p. 243). This suggests that when you say “I reasoned that X,” you might be engaging in what Doris (2015) calls “post-hoc confabulatory storytelling,” a phrase that also describes most academic conference panels (p. 67).
Practical implications proliferate like philosophical rabbits. Ethics and education demand that we distinguish thinking from reasoning: critical-thinking curricula teach students how to reason properly (which assumes reasoning can be taught, an optimistic proposition), while moral psychology investigates how people actually think about right and wrong (which often reveals that they don’t reason so much as emote and rationalize afterward; see Haidt, 2001, p. 814). In artificial intelligence, we ask whether machines “think” (the Turing test’s ambiguous legacy) or simply perform “reasoning” (symbolic computation or neural pattern-recognition that might or might not count as thinking, depending on who’s funding the grant). Legal ethics guidelines now warn lawyers to “supplement rather than replace their own reasoning” with AI tools, an assumption that lawyers were reasoning in the first place (American Bar Association, 2024, p. 12).
Outstanding problems remain, like unsolved puzzles passed down through generations of graduate students. These include integrating normative and descriptive accounts (bridging the gap between how people ought to reason and how they actually do, a gap roughly the size of the Grand Canyon), clarifying the nature of nonpropositional thought (can you reason without language? Can a dog reason? Can a philosophy professor?), understanding unconscious or implicit reasoning (if you reason without knowing it, did you really reason?), and reconciling continental phenomenological views (thinking as “lived experience” involving angst, coffee, and existential dread) with analytic conceptions (thinking as symbol manipulation involving angst, coffee, and technical jargon).
- Definitions and Historical Background
Conceptual Definitions (Or: How to Sound Smart at Parties)
Philosophers and psychologists, being constitutionally incapable of agreeing on anything except perhaps the existence of chairs (and even that is debatable), have offered various formal definitions that overlap as neatly as Venn diagrams drawn by a drunk octopus.
Holyoak and Morrison (2012), in their magisterial Oxford Handbook of Thinking and Reasoning, a volume so heavy it doubles as exercise equipment, define thinking as “the systematic transformation of mental representations of knowledge to characterize actual or possible states of the world, often in service of goals” (p. 3). Let us parse this carefully, as we would a particularly dense fruitcake. “Systematic” implies that thinking has structure; it is not merely random neuronal firing, though critics of certain philosophical traditions might dispute this. “Transformation” suggests that thinking changes representations, like a mental kitchen blender. “Mental representations” are the contents being blended; they could be images, concepts, sentences in Mentalese, or vague feelings that something is off (like when you forget your anniversary). “Characterize actual or possible states of the world” means thinking can represent reality (actual) or fantasy (possible), hence distinguishing thinking from mere random association, though what counts as a “goal” (survival? reproduction? finishing this paragraph?) remains conveniently vague.
Crucially, thinking, as defined here, need not be rational; it can include daydreaming about revenge, creative leaps that defy logic, and the kind of wishful thinking that convinces people that “this time” the diet will work. Holyoak and Morrison (2012) explicitly note that thinking “need not be optimal or even particularly ‘smart’” (p. 4), a concession that helps explain both the success of reality television and the persistence of certain philosophical positions.
By contrast, reasoning is typically restricted to inferential processes. Holyoak and Morrison (2012) note that reasoning “places emphasis on the process of drawing inferences (conclusions) from some initial information (premises). In standard logic, an inference is deductive if the truth of the premises guarantees the truth of the conclusion… If the truth of the premises renders the truth of the conclusion more credible, but does not bestow certainty, the inference is called inductive” (p. 5). Abduction, or inference to the best explanation, is added for completeness, though proponents of Bayesian approaches tend to argue that it’s really just probabilistic reasoning in a trench coat (Douven, 2017, p. 3).
In short: reasoning = drawing conclusions from premises (like a mathematician or a lawyer billing hours). Thinking = any systematic cognitive transformation (like a novelist, a mad scientist, or someone trying to remember where they parked).
Other sources echo this distinction, though with varying degrees of enthusiasm. The Stanford Encyclopedia of Philosophy entry on Thought notes that reasoning is a type of thinking: “Thinking includes… judging; reasoning (drawing conclusions from premises or evidence)” (Moser, 2020, para. 2). In everyday conceptual hierarchies (Rips & Conrad, 1989, p. 192), people treat thinking as the most general mental category, with reasoning as a specialized subtype, like “mammal” versus “primate,” except that both thinking and reasoning can be performed by creatures who fail to grasp basic syllogisms (i.e., most humans).
Historical Treatments: A Parade of Dead White Men (and a Few Others)
Ancient philosophers, bless their sandaled feet, already tangled with these distinctions. Plato distinguished nous (intuitive intellect, the kind that just gets things) from dianoia (discursive reasoning, the kind that proceeds step-by-step like a drunk person navigating stairs). Aristotle gave us the syllogism and thereby doomed generations of students to memorize “All men are mortal” as if that were the most interesting thing about mortality (Hintikka, 2004, p. 12). Medieval scholars like Ockham (14th century) analyzed thinking in terms of “mental language” sentences, a precursor of Fodor’s language-of-thought hypothesis that suggests medieval philosophers anticipated 20th-century cognitive science by about 600 years, which either says something about medieval brilliance or something about cognitive science’s originality (Panaccio, 2017, p. 45).
The modern period saw rationalists (Descartes, Leibniz) viewing thought as inner speech or as clear and distinct ideas, because nothing says “clarity” like proving God’s existence from the fact that you can imagine a perfect being (Descartes, 1641/1984, p. 35). Empiricists (Locke, Hume) emphasized imagination and association, treating ideas as copies of impressions, chained together by habit and the fact that you’ve seen bread nourish before (Hume, 1748/2007, p. 32). Hobbes (1651), in a moment of mechanical reductionism that would make contemporary neuroscientists blush, wrote: “Reasoning is but reckoning” (Leviathan, Part I, Chapter 5, para. 2). By “reckoning,” Hobbes meant computation, adding and subtracting ideas. This made him either a genius (if you like computational theories of mind) or a philistine (if you think reasoning involves something more soulful than arithmetic).
Kant (1781/1998) divided cognition into understanding (conceptual, rule-governed) and sensibility (intuition), maintaining that logic provides normative rules of thought: “Pure general logic tells us how reason ought to proceed” (p. A53/B77). This normative emphasis, the idea that logic prescribes rather than describes, became a hallmark of the philosophical approach to reasoning. Frege (1893/2013) went further, declaring logic a “normative science” that provides laws for correct thinking (p. xv). This stands in stark contrast to the psychological approach, which asks not how you should think but how you actually think, a distinction that has fueled academic turf wars for over a century.
In the 19th and 20th centuries, formal logic and experimental psychology diverged like estranged siblings who shared a childhood but now refuse to speak at family gatherings. Frege and Boole developed symbolic logic as a normative grammar of thought, while Wundt and James studied “thinking” in experimental psychology laboratories. Wundt used controlled introspection (which proved about as reliable as a weather forecast from a groundhog), and James offered insightful observations that remain quotable to this day (“The art of being wise is the art of knowing what to overlook,” which James, 1890, p. 147, probably thought while overlooking something important).
Behaviorism (Watson, Skinner) then rejected introspection entirely, reducing talk of thinking to behavioristic terms. This phase, much like adolescence, was characterized by rebellion against one’s parents and a tendency to say embarrassing things in public. Watson (1913) famously declared that introspection “has no place in psychological method” (p. 176), a position so extreme that it would later be rejected by his intellectual grandchildren, as is often the case in academic lineages.
The cognitive revolution (1950s-1960s) rehabilitated mental processes with a vengeance, as a prodigal son returning with computational models. Newell and Simon (1972) modeled problem-solving (thinking) algorithmically, showing that computers could simulate human thought, provided you defined “thought” narrowly enough to fit a flowchart (pp. 78-82). Chomsky (1965) and Fodor (1975) proposed an internal “language of thought” (Mentalese) for thinking, especially reasoning, treating cognition as symbol manipulation (Fodor, 1975, p. 78). This made thinking look suspiciously like running a computer program in your head, a view that either vindicates AI or reduces human dignity, depending on your perspective.
In continental philosophy, phenomenologists (Husserl, Heidegger) analyzed noesis (the act of thinking) versus noema (the content thought about), treating thinking as “lived intentionality,” a phrase that sounds profound and probably requires black turtlenecks to fully appreciate (Husserl, 1913/1983, p. 205). This approach, emphasizing first-person experience, stands as far as possible from Anglo-American logicians’ focus on formal structure while still remaining within the same academic discipline.
More recently, cognitive and neuroscientists have mapped brain networks for reasoning (the lateral prefrontal cortex) versus more diffuse networks for spontaneous thought (the default-mode system, the “mind-wandering” network that activates when you are supposed to be paying attention in meetings; Raichle et al., 2001, p. 677). The finding? Your brain has a dedicated system for daydreaming. This explains about 80% of human productivity problems.
What Pertains to Thinking vs. Reasoning
Thinking is the broadest cognitive activity. It involves the systematic transformation of mental representations of knowledge to model actual or possible states of the world, often in service of goals. This includes not only logical problem-solving but also imagining, daydreaming, planning, remembering, and forming beliefs.
Reasoning is a subset of thinking. It focuses specifically on inferential processes, drawing conclusions (deductive, inductive, or abductive) from given information (premises). Reasoning is what happens when you consciously move from evidence to a conclusion.
Examples
Thinking includes daydreaming about a vacation, recalling an embarrassing memory, mentally rotating a shape, having a creative “Eureka!” moment in the shower, making a quick moral judgment (“that feels wrong”), planning a trip, or wondering if you locked the front door.
Reasoning includes solving a logic puzzle (e.g., “All A are B, some B are C, therefore…”), deducing that the butler did it because the candlestick is missing, applying Bayes’ rule (even if rarely used in real life), or constructing a legal argument (“my client is innocent because the glove doesn’t fit”).
Normative Status (How One Should Think)
Thinking is primarily descriptive; it is not inherently governed by logical norms. It can be irrational, biased, creative, or just plain stupid. There is no “wrong” way to daydream.
Reasoning is traditionally normative. Logic, probability theory, and decision theory prescribe how one ought to reason. Frege called logic a “normative science” that tells us the laws of correct thinking. Kant said logic tells us how reason should proceed. Normative standards (validity, soundness, coherence) apply to reasoning.
Descriptive Aspect (How People Actually Think)
Psychologists study thinking empirically: how people solve problems, form concepts, use imagery, make decisions, and fall prey to cognitive biases. First-person reports (e.g., “I am thinking about my grandmother”) are used but are often unreliable as indicators of underlying processes.
Reasoning is studied both normatively (in logic texts) and descriptively (in heuristics-and-biases research). Empirical studies show that human reasoning systematically deviates from logical and probabilistic norms. People commit fallacies, ignore base rates, and are influenced by framing. This gap between normative ideals and actual performance is a central finding.
Typical Methods Used to Study Each
Thinking is investigated through cognitive experiments (problem-solving tasks, memory and imagery tasks, and neuroimaging of creative thought), introspection (though debated), phenomenological reports, and naturalistic observation (e.g., asking people to “think aloud”).
Reasoning is studied using formal methods (proofs, logic, and probability theory) for normative models; behavioral experiments (syllogisms, the Wason selection task, and base-rate problems) for descriptive performance; and computational modeling (rule-based systems or connectionist networks) to simulate reasoning processes.
2. Discipline-Specific Distinctions
Philosophy of Mind and Psychology: The Heavyweights
In philosophy of mind, thinking is often treated as any conscious or unconscious cognitive activity; a definition so broad it could include your toaster if the toaster had beliefs (which it doesn’t, probably). Thinking is tied to intentional mental states: beliefs, desires, imaginings, hopes, fears, and the vague sense that you have forgotten something important (Crane, 2015, p. 45).
Phenomenologists (e.g., Husserl, 1913/1983) distinguish the act of thinking (noesis) from its content (noema). Once understood, the distinction can feel like enlightenment, until you try to explain it to someone and realize you sound like a fortune cookie (p. 205). Analytic philosophers debate whether “I think X” is self-evident (Descartes’ infallible cogito: “I think, therefore I am,” the most famous non-sequitur in history, since “I think” already presupposes existence) or transparent/expressive (Evans, 1982: to know what you think, consider the world, not your mind; p. 225).
In psychology, thinking is studied broadly: concept formation (how do you know a chair when you see one?), mental imagery (can you picture a triangle with four sides? No? Then you are normal), decision-making (why do people choose the worst option while confidently believing it is better?), and problem-solving (how do you get the peanut out of the tube without using the tools? Gestalt psychologists want to know). Stanovich (2009) notes that thinking includes even poorly rational processes: “thinking can be sheer stupidity” (p. 14), a finding that explains both internet comments and certain published philosophical papers.
By contrast, in epistemology and logic, reasoning takes center stage like a diva demanding the spotlight. Reasoning concerns propositional attitudes and inference: how beliefs are justified (or rationalized), how one arrives at conclusions (validly or invalidly), and what distinguishes good arguments from bad ones (Hintikka, 2004, p. 56). Theories of knowledge analyze reasoning norms (e.g., logic, probability) and the truth-conditions of conclusions. Logic (from Aristotle through modern symbolic logic) isolates rules of valid inference; from this perspective, reasoning is normative, telling you what you should conclude, not what you will conclude (MacFarlane, 2004, p. 24).
Some philosophers (Frege, Kant) argue logic provides necessary rules for any correct thinking, a view that makes logic the grammar of thought, violations of which constitute not just error but perhaps sin (though Frege stopped short of damnation). As MacFarlane (2004) summarizes: “Frege famously insisted that logic is a normative science, a science of the laws of correct thinking” (p. 24). This normative stance has been challenged by those who think logic describes actual cognitive processes (a view called “psychologism,” which Frege, 1893/2013, p. xv, attacked with the ferocity of a man defending his life’s work).
Cognitive Science and Neuroscience: The New Kids on the Block
Cognitive scientists often use “thinking” as a catch-all for mental operations, a convenient garbage-can category for anything that happens in the head that is not breathing or digesting. They study various subdomains (memory, imagery, decision-making, problem-solving, attention, and consciousness, basically everything except neuroanatomy, which is someone else’s problem; Thagard, 2019, p. 12).
For example, experiments measure the speed of mental imagery in visual rotation tasks (Shepard & Metzler, 1971). Participants mentally rotate three-dimensional shapes, and their response time increases linearly with the rotation angle, suggesting that thinking with images is analog rather than digital and that your brain is secretly a physics simulator (Kosslyn, 1994, p. 45). Kosslyn, Thompson, and Ganis (2006) summarize decades of research showing that “visual mental imagery engages much of the same neural machinery as visual perception” (p. 14), a finding that implies your imagination is essentially perception without sensory input, like a movie theater with no screen.
Meanwhile, reasoning has its own research program: experiments on syllogisms (all A are B; some B are C; therefore…?), conditional reasoning (if P then Q; P; therefore Q, unless you are a psychologist studying human performance, in which case the correct answer is “it depends on how you frame the question”), and probabilistic judgment (how likely is it that Linda is a feminist bank teller? Tversky & Kahneman, 1983, p. 297, found that people systematically violate the conjunction rule).
Here, researchers distinguish normative models (formal logic, Bayesian inference, the “ideal” standards) from “empirical performance” (heuristic biases, the “actual” responses). This distinction has generated more academic publications than any sane person would consider necessary, all of which document that humans are irrational in predictably stupid ways (Kahneman, 2011, p. 12).
Brain imaging reveals partially overlapping yet dissociable networks: lateral prefrontal regions activate during effortful logical reasoning (such as solving syllogisms or deciding whether to finish this paragraph), whereas default-mode regions activate during spontaneous thought or imagery (such as daydreaming about lunch or planning your escape from this reading; Raichle et al., 2001, p. 677). This neurological division mirrors the psychological distinction between “System 2” (slow, deliberate, effortful reasoning) and “System 1” (fast, automatic, effortless thinking), a dual-process theory that has become dominant in cognitive psychology (Evans & Stanovich, 2013, p. 223).
In sum, thinking encompasses consciousness and cognition more generally, whereas reasoning in philosophy and logic is a specialized, normative category (the spicy part). However, in practice, they overlap messily, like Venn diagrams drawn by someone with Parkinson’s. Solving a complex problem (thinking) often involves drawing logical inferences (reasoning), and creative leaps (imagination) can be guided by underlying patterns that may or may not count as reasoning, depending on how generously you use the term.
3. Normative vs. Descriptive Perspectives: The Battle for Your Mind
A key axis, perhaps the key axis, in this discussion is how thinking or reasoning ought to occur versus how it does occur. This is the normative-descriptive distinction, and it has generated enough academic controversy to fill several libraries and ruin countless dinner parties.
Normative Accounts: How You Should Think (If You’re Perfect)
In a normative sense, we ask: What is the ideal standard for reasoning? Classical logic, probability theory, decision theory, and rational choice theory all articulate normative rules for inference and belief updating. These are the “shoulds” of cognition: the standards against which actual thinking is judged and usually found wanting.
Kant (1781/1998) argued that “pure general logic tells us how reason ought to proceed” (p. A53/B77). For Kant, logic provided the rules for correct thinking, not a description of how people actually think (since people are messy, emotional, and easily distracted), but a prescription for how they should think if they want to be rational. This is like a dietitian giving you a meal plan: you should eat kale and spinach, but you will eat pizza and feel guilty about it.
Frege (1893/2013) went further, declaring logic a “normative science” that “prescribes the laws of correct thinking” (p. xv). For Frege, logic was not merely descriptive of how people reason (since people reason badly), but prescriptive of how they ought to reason. This distinguishes logic from psychology: psychology describes actual cognitive processes (warts and all), while logic prescribes ideal cognitive processes (smooth and flawless, like a marble statue of reason). As MacFarlane (2004) puts it: “Frege’s insistence that logic is normative was a central plank in his anti-psychologism” (p. 24).
In practice, a logic lesson tells students that an argument is valid only if its conclusion must be true given the premises, not if it happens to be true, not if it feels true, not if you really want it to be true. This is how one ought to reason, regardless of how one actually reasons (which often involves wishful thinking, confirmation bias, and the kind of motivated reasoning that would embarrass a used car salesman).
Similarly, in ethics, one studies moral reasoning according to rational principles: Kantian duty (act only according to maxims that could be universal law, good luck with that), utilitarian cost-benefit (maximize happiness, but whose happiness? And how do you measure it?), and virtue ethics (be a good person, but what counts as good?). All of these are normative: they prescribe how you should reason about moral questions, not how you actually do reason (which often involves gut feelings, emotional reactions, and the kind of post hoc rationalization that makes philosophers weep; Haidt, 2001, p. 814).
In education, “critical thinking” curricula often teach normative reasoning skills: how to evaluate arguments (without being fooled by fallacies), avoid fallacies (such as ad hominem: “You’re wrong because you’re ugly”), apply statistical reasoning (base rates matter, people!), and distinguish correlation from causation (ice cream sales don’t cause drowning; both increase in summer, a fact that escapes approximately 47% of the population). These curricula assume that reasoning can be taught, an optimistic proposition that conflicts with evidence that people’s reasoning abilities are largely stable and that teaching logic has limited transfer (Stanovich, 2009, p. 187).
Descriptive Accounts: How You Actually Think
In contrast, a descriptive approach investigates how people actually think and reason, warts, biases, irrationalities, and all. Cognitive psychology has documented many departures from normativity, like a police blotter for cognitive crimes:
Confirmation bias: seeking evidence that confirms your beliefs while ignoring disconfirming evidence (Nickerson, 1998, p. 175). This helps explain why reading the news feels like tribal warfare and why political arguments never end.
Base-rate neglect: Ignoring statistical base rates in favor of individual case information (Tversky & Kahneman, 1974, p. 1124). Example: “I know smoking causes cancer, but my grandfather smoked until 95 and died in a car crash, so…”
Anchoring: Being influenced by irrelevant numbers (Tversky & Kahneman, 1974, p. 1128). Example: If someone mentions a high number before you guess, you’ll guess higher. This explains both salary negotiations and why $99.99 seems cheaper than $100.
Availability heuristic: Judging probability by how easily examples come to mind (Tversky & Kahneman, 1974, p. 1127). Example: “Plane crashes are scary because I can picture them, but I can’t picture dying of heart disease, so flying must be more dangerous than eating chips.” (It’s not.)
Framing effects: Responding differently to the same information depending on how it’s presented (Tversky & Kahneman, 1981, p. 453). Example: “90% survival rate” sounds better than “10% mortality rate,” even though they are equivalent. This explains why doctors’ word choice affects patient decisions.
Kahneman and Tversky (1979) showed that people systematically violate expected utility theory (the normative standard for rational decision-making) in predictable ways: they are risk-averse for gains but risk-seeking for losses, a pattern captured by “prospect theory” (p. 263). This discovery earned Kahneman a Nobel Prize and demonstrated that descriptive models of actual decision-making differ fundamentally from normative models of rational decision-making.
Stanovich and West (2000) highlight the normative-descriptive gap in reasoning: people often fail to follow logical rules despite knowing them, revealing a gap between competence (normative potential, what you could do if you tried) and performance (what you actually do when you are not trying very hard; p. 645). This is like knowing the rules of chess but still moving your knight in L-shapes that are not actually L-shapes, except instead of chess, it is reasoning about whether to vaccinate your children, and instead of losing the game, you lose public health.
Dual-process theories separate fast, intuitive “thinking” (System 1) from slow, analytical “reasoning” (System 2; Evans & Stanovich, 2013, p. 223). System 1 is automatic, effortless, associative, and prone to bias. It is the part of you that knows 2+2=4 without thinking. System 2 is controlled, effortful, rule-based, and more accurate. It is the part of you that calculates 17×24 and wishes you had a calculator. The problem? System 2 is lazy, like a teenager who could do their homework but would rather scroll through social media. As Kahneman (2011) puts it: “System 2 is the supporting character who thinks she is the star” (p. 45).
Thus, reasoning in the normative sense sets the ought-to-think standards that philosophers and logicians care about. Thinking in the descriptive sense is studied empirically by psychologists: we measure error rates (people are often wrong), reaction times (people are slower when thinking hard), neural correlates (brains light up in interesting patterns), and individual differences (some people are better at reasoning than others, a fact that explains both academic success and why some people believe the earth is flat).
As the Stanford Encyclopedia of Philosophy entry on the normative status of logic notes, logic is traditionally viewed as a “normative science” prescribing our thought processes (MacFarlane, 2004, p. 24). In practice, ethicists, AI designers, and policy-makers strive to close the normative-descriptive gap (improving human reasoning, building “rational” AI) while acknowledging limitations, like the fact that humans are emotional, distracted, and prone to believing things because they want them to be true.
4. First-Person Self-Reports: “I Think” vs. “I Reason” (Or, How to Confabulate Convincingly)
When you say “I think that P,” you are attributing a belief or judgment to yourself. This is a first-person report of a propositional attitude: “I believe P” (or “I suspect P,” “I am pretty sure P,” “I’m leaning toward P but could be convinced otherwise depending on the evidence and my mood”). Philosophers debate how this works, as evidenced by the intensity of arguments over whether pineapple belongs on pizza.
The Transparency View: Look Outside, Not Inside
One influential view (the transparency view) holds that to know “I think P,” you focus on the content P rather than introspecting on a private mental image (Evans, 1982, p. 225). That is, you don’t look inward at your mental states; you look outward at the world and ask, “Is P true?” If you conclude that P is true (based on evidence, argument, or wishful thinking), you thereby know that you think P. As Evans (1982) famously put it: “In making a self-ascription of belief, you are not looking inward at your own mental states, but rather looking outward at the world” (p. 225).
This view has the advantage of parsimony (it does not posit an internal “mind’s eye” scanning of mental content) and the disadvantage of being counterintuitive (it feels as if you are looking inward, even if you are not). As Moran (2001) argues, the transparency of belief means that “the question of what I believe is answered by considering the world, not by considering my own psychology” (p. 62), which either makes perfect sense (if you think about it) or sounds like gaslighting (if you do not).
The transparency view also implies that first-person authority (the special access you have to your own thoughts) comes not from infallible introspection but from transparency: you do not need to introspect because your beliefs are just your take on the world. This explains why you can be wrong about whether you believe something (if you are confused about the evidence) but not about what you believe (since believing just is taking something to be true). Or so the theory goes.
The Cartesian View: Looking Inward Infallibly (Allegedly)
Descartes (1641/1984) famously took “I think” as indubitable, the cogito that grounds all knowledge: “I think, therefore I am” (p. 35). For Descartes, thinking was self-intimating: you cannot be mistaken about whether you are thinking (though you can be mistaken about what you are thinking about if your thoughts are confused). This view, that we have privileged, infallible access to our own thoughts, dominated philosophy for centuries and still influences folk psychology (people think they know their own minds, even when evidence suggests otherwise). Contemporary analytic philosophers generally accept that we have privileged access to our own thoughts (we know them differently from how we know others’ thoughts) but not infallible access (we can be wrong about our own mental states; Gertler, 2012, p. 97). The debate concerns how this privileged access works and how reliable it is.
The Confabulation Problem: You Don’t Know Yourself as Well as You Think
Here is where things get embarrassing. Nisbett and Wilson (1977) delivered a devastating empirical critique of introspection, showing that people are “almost completely unaware” of the actual causes of their judgments and decisions (p. 233). In their classic studies, participants made choices (e.g., which pair of pantyhose to prefer) and then explained their choices. Their explanations cited factors like “quality,” “feel,” or “color,” but the actual determining factor was position (people preferred the rightmost item because that is what they saw last). When asked directly, participants denied being influenced by position, even when position was the only factor (Nisbett & Wilson, 1977, p. 243). This is confabulation: fabricating plausible-sounding reasons for behavior that was actually caused by factors outside conscious awareness. As Nisbett and Wilson (1977) conclude: “People have no direct introspective access to higher-order cognitive processes” (p. 231). Instead, they infer their own mental processes using implicit theories, just as they infer others’ mental processes, and they are often wrong.
Carruthers (2011) argues that this confabulation extends to self-ascriptions of belief and reasoning: when you say “I think P” or “I reasoned that P,” you are engaging in interpretive processes, not introspective ones (p. 45). You are interpreting your own behavior (including verbal behavior and mental imagery) using the same folk-psychological theory you use to interpret others. The difference? You have more data about yourself (including inner speech, imagery, and feelings) but not better access to causality. As Carruthers (2011) puts it: “The mind is opaque to introspection” (p. 67).
“I Think” vs. “I Reason”: A Distinction That May Not Exist in Practice
Saying “I reason this way” is more reflexive than “I think”: it means “this is how I justify or infer” or “this is the process I’m consciously following.” It involves metacognition, or thinking about thinking, which is itself a form of thinking (Flavell, 1979, p. 906). For example, “I reason by examining the evidence step by step” implies a strategy, an awareness of method, and a claim about how you arrived at your conclusion. Both forms involve intentionality: our thoughts are about things. “I think that it will rain” ascribes a mental state with content “it will rain.” “I reason that it will rain because the clouds are dark” specifies not only the content but also the justification (inferential support) and the process (observation → induction → conclusion). This introduces propositional attitudes (belief, hope, doubt, suspicion, uncertainty) and metacognition (awareness of one’s own cognitive processes).
The difference between “I think” and “I reason” reflects a difference in commitment: “I think” reports a mental state; “I reason” reports a method for arriving at that state. However, given Nisbett and Wilson’s (1977) findings, both reports may be confabulated post hoc rationalizations rather than accurate descriptions of actual cognitive processes (p. 231). When you say “I reasoned that P because of Q,” you may be telling a story about yourself that bears the same relation to your actual cognitive processes as campaign promises do to actual governance: hopeful, plausible, and largely fictional.
Nevertheless, first-person ascriptions remain crucial in philosophy and psychology; they show that thinking is at least partly accessible to consciousness, even if the underlying mechanisms often are not. As Gertler (2012) argues, even if introspection is not infallible, it provides prima facie evidence that deserves weight in theorizing (p. 97). The alternative: dismissing all first-person reports as confabulation, leads to eliminativism about consciousness, which is intellectually coherent but practically absurd (since you are conscious of reading this sentence, aren’t you?).
5. Language, Imagery, and Non-Propositional Thought: Can You Think Without Words?
A longstanding debate, what Dennett (1991) called “the most important unresolved question in cognitive science” (p. 217), is whether thinking is essentially linguistic or can be nonlinguistic. Does thinking require language (inner speech, Mentalese, or natural language sentences), or can we think in images, feelings, smells, and the vague sense that something is off?
The Language of Thought Hypothesis
The Language of Thought Hypothesis (LOTH), most forcefully defended by Fodor (1975), holds that cognition is compositional and systematic, like a language, and that thinking involves manipulating mental symbols in a “Mentalese” vocabulary (p. 78). For example, you might have the Mentalese sentences [WHALES ARE MAMMALS] and [MOBY DICK IS A WHALE] and, through inference, arrive at [MOBY DICK IS A MAMMAL]. This is like programming: symbols are combined according to syntactic rules, and the rules are defined over form rather than meaning, which is why computers can manipulate symbols without understanding them (Fodor, 1975, p. 87).
Evidence for LOTH includes the productivity of thought (you can think infinitely many thoughts despite finite brain resources) and systematicity (if you can think “John loves Mary,” you can also think “Mary loves John”; Fodor & Pylyshyn, 1988, p. 12). If thinking is like a language, it has a combinatorial syntax that generates new thoughts from old components, explaining how you can think “The square root of 2 is irrational” even if you have never thought that exact sentence before.
If LOTH is true, reasoning is the manipulation of Mentalese symbols according to rules, which is either beautiful (if you like computational theories of mind) or horrifying (if you think reasoning is more than symbol-shunting). Fodor (1975) embraced the conclusion: “Thinking is computation” (p. 78). This makes AI possible in principle (since computers can compute) and reduces the mind-body problem to “how does meat compute?” Still mysterious, but at least it is a specific kind of mystery.
Mental Imagery: Pictures in the Head?
However, mental imagery research suggests that some thinking is analog rather than digital, like a picture rather than a sentence. Shepard and Metzler (1971) had participants mentally rotate three-dimensional shapes; response times increased linearly with the rotation angle, suggesting that mental rotation is continuous (like physically rotating an object) rather than discrete (like checking logical relationships; p. 702). Kosslyn (1994) showed that mental images have spatial properties: scanning across an image takes longer over greater distances (p. 45), and smaller images are harder to inspect (p. 67).
Kosslyn, Thompson, and Ganis (2006) summarize neuroimaging evidence: “Visual mental imagery engages much of the same neural machinery as visual perception” (p. 14). The primary visual cortex (V1), the brain’s first stop for visual input, is activated during mental imagery, even in the absence of visual input. This suggests that mental images are not merely descriptions but depictions, representation formats that preserve spatial structure (Kosslyn et al., 2006, p. 15).
This poses a problem for LOTH if LOTH claims that all thought is propositional (sentence-like). If mental images are analog representations (preserving spatial relations) rather than digital representations (arbitrary symbols), then thinking involves at least two formats: propositional (for abstract reasoning) and depictive (for spatial reasoning). As Pylyshyn (2003) argues, however, mental images might be epiphenomenal; the conscious experience of imagery might be a side effect of underlying propositional processing, not the processing itself (p. 85). This debate continues, as with most philosophical debates, with neither side convincingly winning.
Nonpropositional Thought: Beyond Language and Images
Nonpropositional thought extends beyond imagery to include:
Procedural knowledge: Knowing how to do things (ride a bike, tie your shoes) rather than knowing that something is true (Ryle, 1949, p. 32). You can ride a bike without being able to describe how you do it, suggesting that the thinking is not propositional.
Spatial navigation: Finding your way through a building using cognitive maps (Tolman, 1948, p. 192). These maps are geometric rather than linguistic; they preserve distance and angle, not logical form (Rescorla, 2014, p. 2).
Emotional appraisals: Knowing that a situation is dangerous without being able to explain why (Damasio, 1994, p. 128). Your body reacts (heart races, palms sweat) before your conscious reasoning kicks in, suggesting that thinking is embodied rather than linguistic.
Intuitive pattern recognition: Recognizing a friend’s face, hearing a familiar tune, or knowing that a chess position is “good” without being able to articulate the reasons (Polanyi, 1966, p. 4). This “tacit knowledge” operates below the level of conscious propositional thought.
Developmental and anthropological evidence also indicates prelinguistic thought: infants solve simple problems (Baillargeon, 1987, p. 122), and preverbal children learn categories visually (Mandler, 2004, p. 45). Even adults use “inner speech” only for complex or self-directed thinking (Vygotsky, 1962, p. 45); for routine tasks, thinking is often imagistic or procedural.
The Pluralist View: Multiple Formats for Multiple Tasks
Most contemporary cognitive scientists adopt a pluralist view: thinking involves multiple representational formats (propositional, depictive, procedural, embodied), each suited to different tasks. Abstract reasoning about logic and mathematics may be propositional; spatial reasoning about layouts and navigation may be depictive; skill learning may be procedural; and emotional appraisal may be embodied. As Clark (2008) argues, the brain is a “heterogeneous computational engine” that uses whatever format works (p. 45).
Reasoning, however, is often defined as inferential, drawing conclusions from premises. This typically requires propositional representations (since premises and conclusions are propositions). But analogical reasoning (e.g., “The atom is like a solar system”) relies on imagery, and spatial reasoning (e.g., “If A is left of B, and B is left of C, then A is left of C”) can proceed through imagery rather than logic (Gentner & Holyoak, 1997, p. 32).
Thus, the relationship between thinking and reasoning is not “thinking = propositional, reasoning = non-propositional” but rather “thinking can be non-propositional, but reasoning typically requires propositional formats, except when it does not.” Clear? Good. Philosophers have been clarifying this for centuries, and they are not done yet.
Methodological Issues and Critiques: How Not to Study Thinking
Studying thinking and reasoning poses challenges that would make even the most patient researcher reach for something stronger than coffee.
The Introspection Problem: You are Probably Wrong About Your Own Mind
Nisbett and Wilson (1977) demonstrated that people have limited access to their own cognitive processes. In their studies, participants reported reasons for their judgments that were demonstrably false (e.g., claiming quality as the reason for choosing the rightmost pantyhose when position was the actual determinant; p. 243). The authors conclude: “There may be little or no direct introspective access to higher-order cognitive processes” (p. 231). This finding, replicated across dozens of studies, has devastating implications for research programs that rely on first-person reports of thinking and reasoning. If people do not know why they think what they think, asking “How did you reason?” may yield confabulation rather than insight. As Wilson (2002) puts it: “People’s introspective reports about the causes of their own behavior are often inaccurate” (p. 101).
This does not mean first-person reports are useless; people can report what they are thinking (e.g., “I’m thinking about my grandmother”) with reasonable accuracy (Ericsson & Simon, 1980, p. 215). But they cannot reliably report how they arrived at that thought; the process of thought may be opaque to consciousness. This distinction between content access (people know what they are thinking) and process access (people do not know how they got there) is crucial (Wilson, 2002, p. 105).
The Conceptual Analysis Problem: Words Are Slippery
Conceptual analysis of “thinking” versus “reasoning” also has limits. Anglo-American philosophy often relies on linguistic intuitions: “Would we say X counts as thinking?” But critics (Wittgenstein, 1953/2009) question whether such terms have stable definitions across contexts: “The meaning of a word is its use in the language” (p. 43). If people use “thinking” and “reasoning” interchangeably in daily life, a sharp philosophical distinction may be stipulative rather than descriptive.
Cross-linguistic studies show that different languages carve up mental concepts in different ways. German distinguishes Denken (thinking broadly) from Überlegen (deliberate reasoning) from Vernunft (reason as a faculty). Japanese has kangae (thought/idea) and sui-ri (reasoning/logic) with distinct connotations. The assumption that English “thinking” and “reasoning” map neatly onto universal cognitive categories is provincial; a fact philosophers are slowly acknowledging (Machery, 2009, p. 67).
The Normative-Descriptive Confusion: Is/Ought Problem
The normative-descriptive distinction is often blurred in practice. Psychologists sometimes criticize people for violating logical norms, yet why should we expect people to follow them? Logic prescribes how one ought to reason for certain purposes (e.g., mathematical proof, scientific inference), whereas everyday thinking serves different purposes (e.g., speed, social coordination, emotional regulation; Evans, 2002, p. 457).
As Cohen (1981) argued, calling ordinary reasoning “irrational” for violating logical norms is like calling a fish “irrational” for not climbing a tree: you are using the wrong standard (p. 317). Everyday reasoning is ecologically rational, adapted to real-world environments where perfect logic is too slow and Bayesian updating is computationally infeasible (Gigerenzer, 2008, p. 22). The “heuristics and biases” that Kahneman and Tversky identified may be features, not bugs, serving as fast and frugal shortcuts that work well enough in natural environments (Gigerenzer & Gaissmaier, 2011, p. 453).
This critique does not fully let people off the hook; some biases (e.g., base-rate neglect) cause real-world errors in contexts where statistical reasoning matters (e.g., medical diagnosis and legal judgment). But it suggests that the normative-descriptive gap is not merely a gap between “ideal” and “actual” rationality but a gap between different kinds of rationality (logical vs. ecological).
The Neuro-Reductionism Problem: Brains Are Not Theories
Some cognitive scientists argue that folk-psychological terms such as “belief,” “thought,” and “reasoning” lack a one-to-one mapping onto brain processes (Churchland, 1981, p. 67). The brain does not have a “reasoning module” or a “thinking area”; it has distributed networks that support various cognitive functions. The folk categories are useful fictions for everyday explanation, not scientifically respectable categories.
This “eliminativist” view suggests that the distinction between thinking and reasoning may dissolve as we understand brain function at the neural level. Just as “phlogiston” and “caloric fluid” disappeared from chemistry, “thinking” and “reasoning” may disappear from neuroscience (Churchland, 1981, p. 70). Most philosophers and psychologists resist this conclusion, arguing that folk psychology is indispensable for explaining human behavior (Dennett, 1987, p. 45), yet the challenge remains.
Methodological Pluralism
Given these problems, the field has adopted methodological pluralism: combining multiple methods (philosophical analysis, behavioral experiments, neuroimaging, and computational modeling) and triangulating across them. No single method has the final word; each has limitations, and convergence across methods is evidence of robustness (Nisbett & Wilson’s critique of introspection does not doom all first-person methods, only naive ones).
Mixed methods (e.g., “neurophenomenology,” which combines first-person reports with neuroimaging; Lutz & Thompson, 2003, p. 45) seek to link subjective experience to brain data, but the gap remains wide. The mind-body problem, how subjective experience relates to physical processes, is not solved by better methods; it is just illustrated more vividly.
7. Practical Implications: Why You Should Care (Besides Passing Your Exam)
Ethics: Thinking About Right and Wrong (And Whether You are Actually Doing It)
The thinking/reasoning distinction plays out in moral psychology. Some theories posit intuitive thinking (empathic, quick, emotional) versus deliberative reasoning (slow, principled, logical) in ethics (Greene, 2010, p. 45). Greene’s dual-process model of moral cognition suggests that spontaneous moral judgments (e.g., “Pushing the fat man off the bridge is wrong”) result from the emotional system (System 1), while utilitarian calculations (e.g., “Pushing him saves five lives, so it’s right”) result from cognitive reasoning (System 2; Greene, 2010, p. 47).
Debates between rationalist and sentimental ethics echo this distinction: should moral reasoning follow universalizable principles (Kant, 1785/1998, p. 15) or arise from innate sentiments (Hume, 1740/1978, p. 457)? Clarifying what people ought to think (moral reasoning) versus how they do think (moral intuition) is central to applied ethics. For example, if most moral judgments are intuitive, moral education might focus on shaping intuitions (through exemplars, narratives, and practice) rather than on teaching ethical principles (Haidt, 2001, p. 814).
Education: Teaching Thinking (If That’s Even Possible)
Critical thinking education explicitly teaches formal reasoning skills (logic, the scientific method, and probability) on the assumption that they improve thinking (Ennis, 1989, p. 4). But educators also emphasize metacognition (“thinking about thinking”) and creativity (divergent thinking), going beyond pure logic.
Understanding that reasoning skills may not automatically transfer to other domains (Stanovich’s “lawyer–engineer” pattern, pp. 187-188) helps design curricula. Knowing how to evaluate syllogisms in a logic class does not mean you will evaluate real-world arguments well; transfer requires training in application, not just abstract principles (Perkins & Salomon, 1989, p. 18). So, teaching “critical thinking” as a separate course may be less effective than embedding it in domain-specific contexts (e.g., teaching reasoning in biology class about evolution, in history class about evidence).
Artificial Intelligence: Do Machines Think, or Do They Just Compute?
AI raises the question that has launched a thousand conference panels: can machines think, or do they only reason (or neither, or both, depending on definitions)?
Classical AI (expert systems, symbolic AI) embodied a reasoning paradigm: symbolic logic, planning algorithms, and rule-based inference. If Fodor’s language-of-thought hypothesis is correct, classical AI is thinking, just running Mentalese on different hardware (Fodor, 1975, p. 78). If thinking requires consciousness (whatever that is), then classical AI is not thinking (since computers are probably not conscious).
Machine learning (neural networks) often operates as “mindless statisticians,” finding patterns in data without symbols, rules, or explicit representations (Marcus, 2018, p. 45). This is thinking if you define it broadly (information processing) but not reasoning if you define it narrowly (inference from explicit premises). Large language models like GPT-4 produce outputs that resemble human reasoning, but it remains unclear whether they are actually reasoning or merely performing sophisticated pattern matching (Bender & Koller, 2020, p. 5185).
The AI safety community distinguishes between optimized reasoning (algorithms that reliably achieve goals) and human cognitive biases (systematic errors). If AI systems inherit human biases from training data, they will amplify them, which is not ideal for systems making medical diagnoses or loan decisions (O’Neil, 2016, p. 45). Modern practice (e.g., legal ethics guidelines) advises humans to augment their reasoning with AI rather than relinquish thinking entirely. The American Bar Association (2024) warns: “Lawyers should supplement rather than replace their own reasoning with AI tools” (p. 12). This is either wise guidance or a losing battle, depending on how good AI gets.
Law: Reasoning About Rules (While Thinking About Loopholes)
Legal reasoning is a domain where normative standards are paramount; judges and juries are supposed to reason correctly (justifiability, consistency, precedent). Lawyers are trained to reason within formal rules (statutes, precedents, procedures) and to spot fallacies in opponents’ arguments. Psychology of law, however, shows that judges and juries engage in “thinking” influenced by emotion, heuristics, and implicit biases (Kahneman, 2011, p. 45). Anchoring affects sentencing (judges give longer sentences after seeing high anchor numbers; Englich, Mussweiler, & Strack, 2006, p. 189). Confirmation bias affects the evaluation of evidence (people interpret evidence to support their initial beliefs; Nickerson, 1998, p. 175). The ideal of pure legal reasoning collides with the reality of human cognitive biases, which is why legal systems have appellate review, procedural safeguards, and (ideally) diverse decision-makers.
The recent ethics guidelines for attorneys (US bar associations) caution that generative AI should only supplement, not replace, a lawyer’s own reasoning (American Bar Association, 2024, p. 12). This underscores a pragmatic rule: AI can perform rapid “thinking” (retrieving cases, summarizing documents, spotting patterns), but final legal reasoning (applying law to facts, assessing relevance, exercising discretion) demands human oversight. For now.
Everyday Life: Think Before You Reason (Or Vice Versa)
The distinction matters informally, too. When someone says “Think!” they often mean “Use your common sense!” (System 1, fast thinking). When they say “Reason carefully!” they mean “Don’t rely on intuition!” (System 2, slow reasoning). Understanding the difference can help individuals recognize when to slow down and reason through a decision (e.g., financial planning, medical choices) rather than rely on intuition (e.g., “This stock feels right”).
Psychotherapists and coaches train clients in metacognitive strategies: noticing one’s own thought patterns (cognitive behavioral therapy: “What evidence do you have for that thought?”) and critically examining arguments (Socratic questioning: “What do you mean by ‘fair’?”). Both approaches require distinguishing thinking (noticing content) from reasoning (evaluating justification).
8. Unresolved Problems and Research Agenda: What We Still Don’t Know (Which Is a Lot)
Despite centuries of study, from Aristotle to AI, many puzzles remain, like unsolved mysteries passed down through generations of graduate students who hoped to solve them but instead ended up teaching intro to philosophy.
Problem 1: Integrating Normative and Descriptive Accounts
How can we reconcile formal ideals (logic, probability) with messy human cognition? One approach: “bounded rationality” (Simon, 1957, p. 198), in which people are rational given their computational limits and available information. Another: “ecological rationality” (Gigerenzer, 2008, p. 22), in which heuristics are rational in natural environments. A third: “dual-process” (Evans & Stanovich, 2013, p. 223), in which System 1 and System 2 cooperate, with System 2 overriding System 1 when necessary (but often not).
Research agenda: develop cognitive architectures that explain why people violate norms in systematic (not random) ways and whether any normative framework can accommodate human biases. Is “heuristics and biases” a catalog of errors or a taxonomy of adaptive shortcuts? Both, probably, depending on context.
Problem 2: Nature of Non-Propositional Thought
While imagery and nonlinguistic thought are well documented, their precise role in reasoning remains debated. Can abstract reasoning (e.g., mathematical proof, scientific inference) be done without symbolic language? Some philosophers say no (Fodor, 1975, p. 78); others say yes (Dennett, 1991, p. 217).
Questions: Do animals or prelinguistic infants “reason” with images? Do aphantasics (people without mental imagery) reason differently? Ongoing work on unsymbolized thinking (Hurlburt, 2011, p. 45) and aphantasia (Zeman, Dewar, & Della Sala, 2015, p. 218) suggests that people think in different ways. Some have vivid imagery, some have none, some have mostly inner speech, and some have abstract unsymbolized thoughts. This variability challenges one-size-fits-all theories.
Problem 3: Conscious vs. Unconscious Reasoning
Many cognitive processes (intuition, pattern recognition, implicit learning) operate outside awareness, yet they influence our conclusions. Philosophically, this raises questions about justification: Can we count unconscious inference as bona fide reasoning? If reasoning requires conscious awareness of premises and steps, then much of what psychologists call “reasoning” (e.g., unconscious Bayesian updating) is not reasoning at all.
Research: elucidate the cognitive neuroscience of insight (sudden “Aha!” solutions) versus deliberate deduction (step-by-step reasoning). Are insights reasoned unconsciously, or are they non-reasoned intuitions? The answer may determine whether we praise or blame people for their “gut feelings.”
Problem 4: First-Person and Social Cognition
How do we attribute thinking and reasoning to others? Theory of mind suggests we use our own thinking patterns as models for others’ minds (Nichols & Stich, 2003, p. 45). But cultural and individual differences in reasoning style (e.g., Eastern dialectical thinking vs. Western analytical thinking; Nisbett, 2003, p. 45) challenge one-size-fits-all accounts.
Future work: cross-cultural studies of reasoning (do people across all cultures commit the same logical errors?) and improved models of metacognition that explain why we overestimate our introspective insight (the “illusion of explanatory depth”; Rozenblit & Keil, 2002, p. 521).
Problem 5: Language and Thought (Again, Still, Forever)
Does language shape the limits of thought? The Sapir-Whorf hypothesis (that language determines thought) is largely discredited, but weaker versions (that language influences thought) have empirical support. For example, speakers of languages that distinguish light blue from dark blue (Russian) or green from blue (Japanese) make faster color discriminations (Winawer et al., 2007, p. 778). Speakers of languages with absolute spatial terms (north/south, not left/right) navigate more effectively in unfamiliar environments (Levinson, 2003, p. 45).
Unresolved: the degree to which multilingualism or linguistic structure affects reasoning processes (not just perception). Do bilinguals reason differently in different languages? Does inner speech structure thought, or merely accompany it? These questions remain open.
Problem 6: Computational Modeling and AI
AI offers new testbeds for theories of thought and reasoning. Can we build machines that mimic human “thinking” (creativity, imagination, daydreaming) or only “reasoning” (logic, probability, planning)? The success of generative models (GPT-4, DALL-E, etc.) raises philosophical questions about understanding and intentionality.
Searle’s (1980) Chinese Room argument challenges strong AI: syntax (symbol manipulation) does not entail semantics (meaning and understanding; p. 417). Language models may simulate reasoning without genuinely reasoning. How would we tell the difference? The Turing test (Turing, 1950, p. 433) is not decisive: passing the test shows that behavior is indistinguishable from human reasoning, but it does not prove that internal states match.
Research challenge: identify principled differences between algorithmic reasoning and human thinking (maybe invoking integrated information Tononi, 2008, p. 216, or embodiment Clark, 2008, p. 45). Without such differences, we may have to accept that machines do think and reason, just not consciously (whatever that means).
Problem 7: Ethical Implications of Cognitive Differences
How should insights into thinking and reasoning inform social policy? If people systematically mis-reason (e.g., base-rate neglect in medical decisions), can we “nudge” them ethically (Thaler & Sunstein, 2008, p. 6)? Libertarian paternalism advocates choice architecture that preserves freedom while improving outcomes, e.g., default enrollment in retirement plans (because people are bad at saving) and opt-out organ donation (because people are bad at registering).
The ethical dimension of reasoning itself (logic rules, rationality norms) is underexplored. Should people be held morally responsible for “irrational” beliefs? If someone believes a conspiracy theory because of cognitive biases (not malice), are they blameworthy? This intersects with law (culpability, competence to stand trial) and democracy (citizens’ informed reasoning; if voters systematically misreason, what does democratic legitimacy require?).
9. Conclusion
After this lengthy tour through definitions, history, disciplinary squabbles, normative ideals, descriptive failures, first-person confabulations, mental imagery, methodological crises, and practical implications, what can we conclude?
First, the distinction between thinking and reasoning is real but fuzzy, much like the distinction between “walking” and “strolling.” Thinking is the broader category, encompassing all systematic mental transformations (including imagery, daydreaming, planning, and remembering). Reasoning is the narrower category, focusing on inferential processes that draw conclusions from premises (whether deductive, inductive, or abductive).
Second, the distinction carries normative weight in philosophy (reasoning is how you should think; logic provides the rules) and descriptive weight in psychology (thinking is how you actually process information; biases are how you actually err). The gap between the normative and the descriptive is the central puzzle of the field and is likely insoluble by purely armchair methods.
Third, first-person self-reports (“I think,” “I reason”) are reliable for content (people know what they are thinking) but unreliable for process (people do not know how they arrived at their thoughts). This asymmetry explains both the success of introspection in phenomenology and its failure in experimental psychology.
Fourth, thinking is not essentially linguistic; people think in images, feelings, procedures, and spatial representations, not just inner speech or Mentalese. Reasoning typically requires propositional formats (since premises and conclusions are propositions), but some forms of reasoning (analogical, spatial) can proceed non-propositionally.
Fifth, practical implications are everywhere: ethics (intuitive vs. deliberative moral judgment), education (teaching critical thinking), AI (do machines think or just compute?), law (reasoning about rules while watching for loopholes), and everyday life (knowing when to trust your gut and when to slow down).
Finally, the unresolved issues are numerous: merging normative and descriptive perspectives, understanding non-propositional thought, differentiating conscious from unconscious reasoning, explaining cultural differences, modeling the interaction between language and thought, developing thinking machines, and considering ethical concerns.
In short: we have made progress, but we are not done. Thinking and reasoning remain what they have always been: fascinating, frustrating, and fundamentally human (for now). As Descartes (1641/1984) said, “Cogito, ergo sum” (p. 35), I think, therefore I am. But what kind of thinking? Does reasoning qualify? And if I am reasoning poorly, do I really exist?
On that note, dear reader, I think I will stop. (Or should I say: I reason that stopping is appropriate given the length of this document, the diminishing returns of additional analysis, and the fact that you probably stopped reading two sections ago.)
References
American Bar Association. (2024). AI And Attorney Ethics Rules: 50-State Survey. Lawyers and the Legal Process Center. https://www.justia.com/trials-litigation/ai-and-attorney-ethics-rules-50-state-survey/
Baillargeon, R. (1987). Object Permanence in 3½- and 4½-Month-Old Infants. Developmental Psychology, 23(5), 122-135.
Bender, E. M., & Koller, A. (2020). Climbing Towards NLU: On Meaning, Form, and Understanding in the Age of Data. Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, 5185-5198.
Carruthers, P. (2011). The Opacity of Mind: An Integrative Theory Of Self-Knowledge. Oxford University Press.
Chomsky, N. (1965). Aspects of the Theory of Syntax. MIT Press.
Churchland, P. M. (1981). Eliminative Materialism and The Propositional Attitudes. The Journal of Philosophy, 78(2), 67-90.
Clark, A. (2008). Supersizing the Mind: Embodiment, Action, and Cognitive Extension. Oxford University Press.
Cohen, L. J. (1981). Can Human Irrationality Be Experimentally Demonstrated? Behavioral And Brain Sciences, 4(3), 317-331.
Crane, T. (2015). The Mechanical Mind: A Philosophical Introduction to Minds, Machines, and Mental Representation (3rd ed.). Routledge.
Damasio, A. R. (1994). Descartes’ Error: Emotion, Reason, and the Human Brain. Putnam.
Dennett, D. C. (1987). The intentional Stance. MIT Press.
Dennett, D. C. (1991). Consciousness Explained. Little, Brown and Company.
Descartes, R. (1984). Meditations On First Philosophy (J. Cottingham, Trans.). Cambridge University Press. (Original work published 1641)
Doris, J. M. (2015). Talking To Our Selves: Reflection, Ignorance, and Agency. Oxford University Press.
Douven, I. (2017). Abduction. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2017 ed.). https://plato.stanford.edu/entries/abduction/
Englich, B., Mussweiler, T., & Strack, F. (2006). Playing Dice with Criminal Sentences: The Influence of Irrelevant Anchors on Experts’ Judicial Decision Making. Personality And Social Psychology Bulletin, 32(2), 188-200.
Ennis, R. H. (1989). Critical Thinking and Subject Specificity: Clarification and Needed Research. Educational Researcher, 18(3), 4-10.
Ericsson, K. A., & Simon, H. A. (1980). Verbal Reports as Data. Psychological Review, 87(3), 215-251.
Evans, G. (1982). The Varieties of Reference. Oxford University Press.
Evans, J. St. B. T. (2002). Logic and Human Reasoning: An Assessment of the Deduction Paradigm. Psychological Bulletin, 128(6), 457-476.
Evans, J. St. B. T., & Stanovich, K. E. (2013). Dual-Process Theories of Higher Cognition: Advancing the Debate. Perspectives on Psychological Science, 8(3), 223-241.
Flavell, J. H. (1979). Metacognition And Cognitive Monitoring: A New Area of Cognitive-Developmental Inquiry. American Psychologist, 34(10), 906-911.
Fodor, J. A. (1975). The language of Thought. Harvard University Press.
Fodor, J. A., & Pylyshyn, Z. W. (1988). Connectionism and Cognitive Architecture: A Critical Analysis. Cognition, 28(1-2), 3-71.
Frege, G. (2013). Basic Laws of Arithmetic (P. A. Ebert & M. Rossberg, Trans.). Oxford University Press. (Original work published 1893)
Gentner, D., & Holyoak, K. J. (1997). Reasoning and Learning by Analogy: Introduction. American Psychologist, 52(1), 32-34.
Gertler, B. (2012). Self-knowledge. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Winter 2012 ed.). https://plato.stanford.edu/entries/self-knowledge/
Gigerenzer, G. (2008). Rationality For Mortals: How People Cope with Uncertainty. Oxford University Press.
Gigerenzer, G., & Gaissmaier, W. (2011). Heuristic Decision Making. Annual Review of Psychology, 62, 451-482.
Greene, J. D. (2010). Moral Tribes: Emotion, Reason, and the Gap Between Us and Them. Penguin Press.
Haidt, J. (2001). The Emotional Dog and Its Rational Tail: A Social Intuitionist Approach to Moral Judgment. Psychological Review, 108(4), 814-834.
Hintikka, J. (2004). Analyses of Aristotle (Vol. 1). Kluwer Academic Publishers.
Hobbes, T. (1651). Leviathan. Andrew Crooke.
Holyoak, K. J., & Morrison, R. G. (Eds.). (2012). The Oxford Handbook of Thinking and Reasoning. Oxford University Press.
Hume, D. (1978). A Treatise Of Human Nature. Oxford University Press. (Original work published 1740)
Hume, D. (2007). An Inquiry Concerning Human Understanding. Oxford University Press. (Original work published 1748)
Hurlburt, R. T. (2011). Investigating Pristine Inner Experience: Moments of Truth. Cambridge University Press.
Husserl, E. (1983). Ideas Pertaining to A Pure Phenomenology and to A Phenomenological Philosophy (F. Kersten, Trans.). Martinus Nijhoff. (Original work published 1913)
James, W. (1890). The Principles of Psychology (Vol. 1). Henry Holt and Company.
Kahneman, D. (2011). Thinking, Fast and Slow. Farrar, Straus and Giroux.
Kahneman, D., & Tversky, A. (1979). Prospect Theory: An Analysis of Decision Under Risk. Econometrica, 47(2), 263-291.
Kant, I. (1998). Critique of Pure Reason (P. Guyer & A. W. Wood, Trans.). Cambridge University Press. (Original work published 1781)
Kant, I. (1998). Groundwork of The Metaphysics of Morals (M. Gregor, Trans.). Cambridge University Press. (Original work published 1785)
Kosslyn, S. M. (1994). Image And Brain: The Resolution of the Imagery Debate. MIT Press.
Kosslyn, S. M., Thompson, W. L., & Ganis, G. (2006). The Case for Mental Imagery. Oxford University Press.
Levinson, S. C. (2003). Space In Language and Cognition: Explorations in Cognitive Diversity. Cambridge University Press.
Lutz, A., & Thompson, E. (2003). Neurophenomenology: Integrating Subjective Experience and Brain Dynamics in the Neuroscience of Consciousness. Journal of Consciousness Studies, 10(9-10), 45-65.
MacFarlane, J. (2004). Frege, Kant, and the Logic of Logic. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2004 ed.). https://plato.stanford.edu/entries/logic-normative/
Machery, E. (2009). Doing Without Concepts. Oxford University Press.
Mandler, J. M. (2004). The Foundations of Mind: Origins of Conceptual Thought. Oxford University Press.
Marcus, G. (2018). Rebooting AI: Building Artificial Intelligence We Can Trust. Pantheon.
Moran, R. (2001). Authority and Estrangement: An Essay on Self-Knowledge. Princeton University Press.
Moser, P. K. (2020). Thought. In E. N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Summer 2020 ed.). https://plato.stanford.edu/entries/thought/
Newell, A., & Simon, H. A. (1972). Human Problem Solving. Prentice-Hall.
Nichols, S., & Stich, S. (2003). Mindreading: An Integrated Account of Pretense, Self-Awareness, and Understanding Other Minds. Oxford University Press.
Nickerson, R. S. (1998). Confirmation Bias: A Ubiquitous Phenomenon in Many Guises. Review of General Psychology, 2(2), 175-220.
Nisbett, R. E. (2003). The Geography of Thought: How Asians and Westerners Think Differently… and Why. Free Press.
Nisbett, R. E., & Wilson, T. D. (1977). Telling More Than We Can Know: Verbal Reports on Mental Processes. Psychological Review, 84(3), 231-259.
O’Neil, C. (2016). Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. Crown.
Panaccio, C. (2017). Ockham’s Nominalism and The Language of Thought. Vivarium, 55(1-3), 45-63.
Perkins, D. N., & Salomon, G. (1989). Are Cognitive Skills Context-Bound? Educational Researcher, 18(1), 16-25.
Polanyi, M. (1966). The Tacit Dimension. Doubleday.
Pylyshyn, Z. W. (2003). Seeing and Visualizing: It’s Not What You Think. MIT Press.
Raichle, M. E., MacLeod, A. M., Snyder, A. Z., Powers, W. J., Gusnard, D. A., & Shulman, G. L. (2001). A Default Mode of Brain Function. Proceedings of the National Academy of Sciences, 98(2), 676-682.
Rescorla, M. (2014). Cognitive Maps and the Language of Thought. UCLA Philosophy Working Paper Series. https://philosophy.ucla.edu/wp-content/uploads/2016/08/Cognitive-Maps.pdf
Rips, L. J., & Conrad, F. G. (1989). Folk Psychology of Mental Activities. Psychological Review, 96(2), 187-207.
Rozenblit, L., & Keil, F. (2002). The Misunderstood Limits of Folk Science: An Illusion of Explanatory Depth. Cognitive Science, 26(5), 521-562.
Ryle, G. (1949). The Concept Of Mind. Hutchinson.
Searle, J. R. (1980). Minds, Brains, and Programs. Behavioral and Brain Sciences, 3(3), 417-457.
Shepard, R. N., & Metzler, J. (1971). Mental Rotation of Three-Dimensional Objects. Science, 171(3972), 701-703.
Simon, H. A. (1957). Models of Man: Social and Rational. John Wiley and Sons.
Stanovich, K. E. (2009). What Intelligence Tests Miss: The Psychology of Rational Thought. Yale University Press.
Stanovich, K. E., & West, R. F. (2000). Individual Differences in Reasoning: Implications for The Rationality Debate? Behavioral and Brain Sciences, 23(5), 645-665.
Thaler, R. H., & Sunstein, C. R. (2008). Nudge: Improving Decisions About Health, Wealth, and Happiness. Yale University Press.
Thagard, P. (2019). Mind-Society: From Brains to Social Sciences. Oxford University Press.
Tolman, E. C. (1948). Cognitive Maps in Rats and Men. Psychological Review, 55(4), 189-208.
Tononi, G. (2008). Consciousness as Integrated Information: A Provisional Manifesto. The Biological Bulletin, 215(3), 216-242.
Turing, A. M. (1950). Computing Machinery and Intelligence. Mind, 59(236), 433-460.
Tversky, A., & Kahneman, D. (1974). Judgment Under Uncertainty: Heuristics and Biases. Science, 185(4157), 1124-1131.
Tversky, A., & Kahneman, D. (1981). The Framing of Decisions and The Psychology of Choice. Science, 211(4481), 453-458.
Tversky, A., & Kahneman, D. (1983). Extensional Versus Intuitive Reasoning: The Conjunction Fallacy in Probability Judgment. Psychological Review, 90(4), 293-315.
Vygotsky, L. S. (1962). Thought And Language (E. Hanfmann & G. Vakar, Trans.). MIT Press.
Watson, J. B. (1913). Psychology as the Behaviorist Views It. Psychological Review, 20(2), 158-177.
Wilson, T. D. (2002). Strangers To Ourselves: Discovering the Adaptive Unconscious. Harvard University Press.
Winawer, J., Witthoft, N., Frank, M. C., Wu, L., Wade, A. R., & Boroditsky, L. (2007). Russian Blues Reveal Effects of Language on Color Discrimination. Proceedings of the National Academy of Sciences, 104(19), 7780-7785.
Wittgenstein, L. (2009). Philosophical Investigations (G. E. M. Anscombe, P. M. S. Hacker, & J. Schulte, Trans.; 4th ed.). Wiley-Blackwell. (Original work published 1953)
Zeman, A., Dewar, M., & Della Sala, S. (2015). Lives Without Imagery: Congenital Aphantasia. Cortex, 73, 218-228.
