Imagination

Abstract

Imagination is a form of cognition: the capacity to represent objects, events, and scenarios that are not present to the senses, whether they are remembered, anticipated, or wholly invented. It spans the vivid picture of an absent friend's face, the spatial manipulation of a shape turned over in the mind, and the construction of a future event that has never occurred. This article defines imagination and its relation to perception and memory, surveys its MeSH subtypes, works through the century-long debate over whether mental images are picture-like or description-like, examines how imagery vividness is measured and why some minds report none at all, links imagination to the constructive memory system that also simulates the future, and derives the reaction-time signature that first made mental imagery experimentally tractable.

Keywords: mental imagery, mental rotation, aphantasia, constructive episodic simulation

Ask someone to picture an apple and most will report an experience: a rounded red shape, perhaps a stem, hanging somewhere in an inner visual field. Ask them to say whether a rotated letter is a normal R or its mirror image and they will feel themselves turning it in the mind before answering. These acts — conjuring the absent, transforming it, projecting it forward in time — are the province of imagination. What makes imagination a scientific subject rather than a poetic one is that these inner acts have measurable structure: they take time in proportion to the transformation performed, they recruit the same visual machinery as seeing, and they vary so sharply between people that some report a mind's eye as vivid as sight while others report no image at all. This article treats imagination as the cognitive system that generates and manipulates internal representations, and follows the evidence from the first questionnaire on imagery to the neuroscience of the constructive brain.

What Imagination Is

Imagination is the cognitive capacity to form and operate on representations of objects, events, or scenarios that are not currently being perceived. MeSH glosses the descriptor tersely, as a new pattern of perceptual or ideational material derived from past experience, and the phrase captures two essentials: imagination is generative, assembling patterns rather than merely replaying them, and it is derived, built from the residue of prior experience rather than from nothing. The most studied form is visual mental imagery, the experience of seeing with the mind's eye, but imagination is not confined to vision: it encompasses auditory imagery, motor imagery, and the abstract simulation of events and their consequences.

Two boundaries help fix the concept. Imagination is distinct from perception in that it operates without a corresponding external stimulus — the signal is generated internally, top-down, rather than driven by the senses — yet the two share representational machinery, which is why an imagined scene can be confused with a remembered percept. It is distinct from memory in that it need not be faithful to any actual event; imagination recombines the elements of experience into arrangements that never occurred. But the boundary with memory is porous, because remembering is itself reconstructive, and the same system that reassembles a past episode can assemble a hypothetical one. These overlaps — with perception on one side and memory on the other — organise the empirical literature and structure the sections that follow.

Types of Imagination

The Medical Subject Headings vocabulary places Imagination (D007092) within cognition and files two narrower descriptors beneath it. These subtypes are not a psychological taxonomy of imagination — MeSH is an indexing classification built to organise the biomedical literature, not a theory of mental faculties — but they mark the two forms of imaginative activity the literature most often isolates, and they are worth naming because each is a large topic in its own right.

The two subtypes are orthogonal along the dimension that matters most for imagination as a whole: volitional control. Dreams are the paradigm of imagery that arises without an act of will and is mistaken for perception; fantasy is the paradigm of imagery deliberately summoned and tagged as unreal. The bulk of experimental work on imagination — mental rotation, image scanning, vividness measurement — concerns the voluntary, waking imagery that fantasy exemplifies, and it is that literature the rest of this article addresses.

The Imagery Debate

The modern science of imagination begins with a demonstration that mental images have measurable structure. Shepard and Metzler showed observers pairs of perspective drawings of three-dimensional block figures and asked whether the two were the same object rotated, or mirror images. The time to decide rose linearly with the angular difference between the two figures, exactly as if observers were mentally rotating one figure into alignment with the other at a roughly constant rate (Shepard & Metzler, 1971). The linearity is the crucial result: it implies the mental operation passes through intermediate orientations, which is what one would expect of a representation that preserves the analog spatial geometry of the object rather than an abstract description of it.

This finding launched a debate that structured the field for three decades. On one side, Kosslyn argued for a depictive account: mental images are quasi-pictorial representations that depict spatial information in a functional space, and manipulating them is subject to the constraints of that space. Image scanning made this concrete: the time to shift attention between two points on a memorised map rises linearly with the distance between them on the imagined image, as though the mind were traversing a real spatial layout (Kosslyn et al., 1978). Neuroimaging strengthened the case by revealing that generating a visual image activates early visual cortex — including primary visual cortex, retinotopically — much as seeing does, suggesting the image is realised in a topographically organised medium (Kosslyn et al., 2001). On the other side, Pylyshyn defended a propositional (descriptionalist) account: the picture-like quality of imagery is how it feels, not how it is represented. The behavioural regularities, he argued, arise from tacit knowledge — people unconsciously simulate what would happen in the corresponding perceptual situation — and cognition is not penetrated by literal pictures in the brain (Pylyshyn, 2003). The demonstration below reproduces the mental-rotation result and lets the reader see the linear relationship between angular disparity and decision time for themselves.

Set the angular disparity between the two figures. The predicted decision time rises linearly with the angle, as if the figure were turned at a constant rate.

The slope is constant: every extra 60° adds about 1000 ms, whatever the starting orientation. That fixed rate is the behavioural fingerprint of an analog spatial representation being rotated, rather than an abstract label being read off.

Measuring the Mind's Eye

Long before the imagery debate, Galton discovered that imagination is not the same for everyone. His 1880 breakfast-table survey asked respondents to picture their morning table and rate the vividness of the image; the replies ranged from perfect, fully coloured scenes to a complete absence of any picture, and some of his most eminent scientific correspondents reported the least imagery (Galton, 1880). Galton had stumbled on a genuine dimension of individual difference and the method — structured self-report of vividness — that would come to measure it.

That method was standardised by Marks, whose Vividness of Visual Imagery Questionnaire (VVIQ) asks respondents to summon specific scenes and rate each on a vividness scale; scores predict performance on tasks that benefit from imagery, giving the self-report criterion validity (Marks, 1973). The extremes of the VVIQ distribution turned out to be scientifically important. Zeman and colleagues described people who report a lifelong, near-total absence of voluntary visual imagery — they know what an apple looks like but cannot see one in the mind — and named the condition aphantasia; the mirror extreme, imagery as vivid as real seeing, they later termed hyperphantasia (Zeman et al., 2015). A large study of both extremes linked them to differences in autobiographical memory, dreaming, and even occupational choice, establishing imagery vividness as a stable trait with real consequences rather than a mere reporting quirk (Zeman et al., 2020). The demonstration below places a self-rated vividness on the population distribution and shows how the same target scene might be represented across the spectrum from aphantasia to hyperphantasia.

Move the slider to set a self-rated imagery vividness. The panel shows how vividly the target — an apple — might appear in the mind's eye, and where that rating falls on the population distribution.

Most people cluster in the middle of the distribution, but the tails are real and stable: aphantasia at the low end reports no voluntary image at all, hyperphantasia at the high end an image as vivid as seeing. The spectrum is a genuine dimension of individual difference, not a difference in how people describe the same experience.

Imagination and Memory

The deepest recent insight into imagination is that it is not a separate faculty but a repurposing of memory. Schacter and colleagues framed the prospective brain: because remembering is a constructive process that reassembles the elements of past episodes, the same machinery can recombine those elements into events that have not happened, which is why episodic memory and episodic future thinking are so tightly linked (Schacter et al., 2007). Neuroimaging bore this out. Remembering a past event and imagining a future one recruit a largely common core network, with the differences concentrated in the greater constructive demand of building a novel event rather than retrieving a stored one (Addis et al., 2007).

The pivotal evidence came from patients. Hassabis and Maguire proposed scene construction — the mental generation and maintenance of a coherent spatial context — as the shared process underlying both episodic memory and imagination, motivated by the finding that patients with hippocampal amnesia are impaired not only at remembering their past but at imagining new experiences, producing fragmented scenes that lack spatial coherence (Hassabis & Maguire, 2007). This common core sits within the default mode network, the set of regions most active when the mind is not engaged with the external world and instead turned to self-generated thought — remembering, planning, imagining, and mental simulation (Buckner et al., 2008). Imagination, on this view, is what the brain's constructive memory system does when it is freed from the constraint of reporting the actual past.

Imagery and Perception

If imagery recruits visual cortex, it should interact with perception, and it does. A synthesis of the evidence holds that visual imagery and visual perception share a substantial part of their neural representation, differing chiefly in the direction of the dominant signal: perception is driven bottom-up by the senses, imagery top-down from memory and knowledge into the same sensory areas (Dijkstra et al., 2019). This shared-representation view explains both why imagery can facilitate or interfere with perception and why the two are sometimes confused.

The overlap also solved a measurement problem. Self-report of vividness is unavoidably subjective, but if imagery uses sensory machinery it should have sensory consequences that can be measured objectively. Pearson and colleagues exploited binocular rivalry: when different images are shown to the two eyes, perception alternates, and briefly imagining one of the patterns beforehand biases which one is subsequently seen — a priming effect proportional to imagery strength. The effect is absent in aphantasia, providing an objective correlate of the self-reported absence of imagery and turning vividness into something measurable without asking (Keogh & Pearson, 2018). A broad review sets these objective methods within the wider cognitive neuroscience of imagery, its mechanisms, and its individual differences (Pearson, 2019). The demonstration below models the binocular-rivalry priming paradigm and shows how imagery strength shifts the probability that the imagined pattern wins the subsequent rivalry.

Choose which pattern to briefly imagine, then set imagery strength. When a red and a green grating are afterwards shown one to each eye, perception alternates — but imagining a pattern first biases which one is seen. Stronger imagery, stronger bias; at zero strength (aphantasia) the priming vanishes.

Because the priming effect scales with imagery strength and disappears when imagery is absent, it turns a private, introspective report into an objective behavioural measure: the mind's eye leaves a measurable trace on what the eyes then see.

Worked Example

The mental-rotation result is quantitative, and its arithmetic makes concrete what analog transformation means. Model decision time as a linear function of angular disparity:

RT(θ) = a + b · θ

Here a is the intercept — the time for encoding, comparison, and response that does not depend on rotation — and b is the cost per degree of rotation, the reciprocal of the mental rotation rate. Take an intercept a = 500 ms and a rotation rate of 60 degrees per second, so that each degree costs b = 1000 / 60 ≈ 16.7 ms. The model then predicts:

RT(60°) = 500 + 16.7 × 60 = 500 + 1002 = 1502 ms

RT(120°) = 500 + 16.7 × 120 = 500 + 2004 = 2504 ms

RT(180°) = 500 + 16.7 × 180 = 500 + 3006 = 3506 ms

The instructive feature is the constant slope. Each additional 60 degrees of disparity adds the same 1002 ms, whatever the starting orientation, because the mind rotates the figure at a fixed angular rate. A representation that stored orientation as an abstract label, such as the figure is at 180 degrees, would predict no such gradient; the decision would take the same time regardless of angle, since comparing two labels is no harder than comparing two others. The linear rise is therefore the behavioural fingerprint of an analog, spatial representation, and it is what Shepard and Metzler found. Figure 1 plots this function.

Discussion

Imagination has proved to be less a single faculty than a convergence of systems the mind evolved for other purposes. Its visual form borrows the machinery of perception, running it top-down; its episodic form borrows the machinery of memory, running it forward in time or sideways into the counterfactual. This borrowing is why imagination is so revealing: to study how someone imagines an apple is to probe the visual system's representations, and to study how they imagine next week is to probe the constructive nature of memory itself. The old intuition that imagination is a faint copy of perception was not wrong so much as incomplete — the copy is faint because it is generated rather than received, but it runs on the same substrate.

The field's central theoretical dispute, the imagery debate, was never cleanly resolved so much as reframed. The demonstration that imagery engages retinotopic visual cortex is hard to reconcile with a purely propositional account, yet the descriptionalist's core caution stands: the felt pictorial quality of an image is not itself evidence of a picture in the head, and tacit knowledge can mimic analog effects. What the debate produced was a more careful science — one that distinguishes the phenomenology of imagery from its representational format, and that increasingly settles such questions with neuroimaging and objective behavioural measures rather than introspection alone.

Current Directions

The discovery of aphantasia has opened the most active current front. Because a sizeable minority of people report little or no voluntary imagery, imagery vividness can now be treated as a natural experiment: comparing aphantasic and typical imagers isolates what imagery actually does in memory, dreaming, face recognition, and emotional response. The objective binocular-rivalry measure and its absence in aphantasia have made the trait tractable without relying solely on introspective report, and work continues to map its perceptual, physiological, and neural profile (Keogh & Pearson, 2018); (Pearson, 2019).

A second front concerns decoding. If imagery and perception share representations, the patterns of activity that reconstruct a seen image should partly reconstruct an imagined one, and advances in neural decoding are beginning to read out the content of imagery and to characterise precisely how imagined and perceived representations differ in strength and locus (Dijkstra et al., 2019). How the top-down signal that generates an image is initiated and controlled — and why its strength varies so widely between people — remains the open question the next decade of imagery research is likely to address.

Common Misconceptions

Everyone experiences mental imagery in the same way.

Imagery vividness varies enormously, from aphantasia (no voluntary imagery) to hyperphantasia (imagery as vivid as seeing), a stable individual difference with measurable consequences (Zeman et al., 2015); (Zeman et al., 2020).

A mental image is literally a picture in the brain.

Whether imagery is depictive or propositional was debated for decades; imagery does engage visual cortex, but the felt pictorial quality is not by itself evidence of a picture-like representation (Kosslyn et al., 2001); (Pylyshyn, 2003).

Imagining the future is unrelated to remembering the past.

The two rely on a common constructive system: hippocampal amnesia impairs both, and imagining and remembering recruit a largely shared brain network (Hassabis & Maguire, 2007); (Schacter et al., 2007).

Glossary

Aphantasia.

A lifelong near-total absence of voluntary visual mental imagery, at the low extreme of the vividness distribution.

Binocular rivalry.

The alternation of perception between two different images presented to the two eyes, used to measure imagery objectively through its priming effect.

Constructive episodic simulation.

The recombination of stored episodic elements into representations of novel or future events by the memory system.

Default mode network.

A set of brain regions most active during self-generated thought, including remembering, planning, and imagining.

Depictive representation.

A quasi-pictorial mental representation that depicts spatial structure in a functional space, as in Kosslyn's account of imagery.

Hyperphantasia.

Visual mental imagery as vivid as actual seeing, at the high extreme of the vividness distribution.

Imagination.

The cognitive capacity to generate and manipulate representations of objects and events not currently perceived.

Mental imagery.

The experience of perception-like representation in the absence of the corresponding external stimulus.

Mental rotation.

The imagined turning of a visual figure, whose duration rises linearly with the angle of rotation.

Propositional representation.

An abstract, language-like encoding of information; on the descriptionalist account, the true format underlying the felt picture of imagery.

Prospective brain.

The proposal that the brain's constructive memory system exists in part to simulate future and hypothetical events.

Scene construction.

The generation and maintenance of a coherent spatial context, proposed as the shared process behind memory and imagination.

Tacit knowledge.

Unconscious knowledge of how a perceptual situation would unfold, invoked to explain imagery effects without a picture-like representation.

Visual mental imagery.

Imagery in the visual modality, the most studied form of imagination and the subject of the imagery debate.

Vividness of Visual Imagery Questionnaire (VVIQ).

Marks's standardised self-report instrument for rating the vividness of summoned visual images.

Key Researchers

Francis Galton (1822-1911). Pioneered the empirical study of mental imagery and individual differences with the 1880 breakfast-table imagery survey. Wikipedia - Wikidata

Stephen M. Kosslyn. Leading proponent of the depictive theory of mental imagery and its grounding in early visual cortex. ORCID - Google Scholar - Wikipedia - Wikidata

Eleanor A. Maguire (1970-2025). Showed that hippocampal amnesic patients are impaired at imagining novel scenes, evidence for scene construction as a shared memory-imagination process. ORCID - Wikipedia - Wikidata

Joel Pearson. Developed objective sensory measures of imagery strength and characterised the perceptual profile of aphantasia. ORCID - Google Scholar

Zenon W. Pylyshyn (1937-2022). Advanced the propositional side of the imagery debate, arguing mental images are not literal pictures in the brain. Wikipedia - Wikidata

Daniel L. Schacter. Developed the constructive-memory and prospective-brain accounts linking imagination to the episodic memory system. ORCID - Google Scholar - Wikipedia - Wikidata

Adam Zeman. Coined the terms aphantasia and hyperphantasia and mapped the psychological significance of imagery-vividness extremes. ORCID - Google Scholar - Wikipedia - Wikidata

Frequently Asked Questions

What is imagination in cognitive psychology? It is the cognitive capacity to generate and manipulate mental representations of objects, events, or scenarios that are not currently being perceived, drawing on the same neural systems used for perception and memory.

Is mental imagery the same as imagination? Visual mental imagery, the experience of seeing with the mind's eye, is the most studied form of imagination, but imagination is broader, encompassing auditory and motor imagery and the abstract simulation of events and their consequences.

What did the mental-rotation experiments show? Shepard and Metzler found that the time to judge whether two figures match rises linearly with the angle between them, implying that mental images preserve analog spatial structure and are transformed at a roughly constant rate (Shepard & Metzler, 1971).

What is the imagery debate? It is the dispute over whether mental images are depictive (picture-like, as Kosslyn argued) or propositional (description-like, as Pylyshyn argued). Imagery engages visual cortex, but whether that settles the question of representational format remains contested (Kosslyn et al., 2001); (Pylyshyn, 2003).

What is aphantasia? Aphantasia is a lifelong near-absence of voluntary visual mental imagery: an aphantasic person knows facts about how things look but does not experience a picture in the mind. It sits at the low extreme of a vividness spectrum whose high extreme is hyperphantasia (Zeman et al., 2015).

How is imagery vividness measured? By self-report, chiefly Marks's Vividness of Visual Imagery Questionnaire, and objectively through sensory consequences such as binocular-rivalry priming, which is absent in aphantasia (Marks, 1973); (Keogh & Pearson, 2018).

Why is imagining the future linked to remembering the past? Both draw on the same constructive memory system, which reassembles stored episodic elements; hippocampal amnesia impairs both remembering and imagining, and the two recruit a largely shared brain network (Hassabis & Maguire, 2007); (Schacter et al., 2007).

Does imagination use the same brain areas as seeing? Substantially yes: imagery and perception share much of their neural representation, differing mainly in whether the signal is driven bottom-up from the senses or top-down from memory and knowledge (Dijkstra et al., 2019).

References