Why inner speech looks like two different things at once
When you silently talk to yourself, two things seem to happen. There is a sound-like quality, as if you hear words without speaking them, and there is a faint sense of movement, as if your speech muscles are preparing but not acting. Researchers have long argued over which of these is the core of inner speech. One view treats it as auditory imagery, a kind of imagined hearing. The other treats it as a language process with motor components, closer to silent articulation.
The brain evidence does not force a choice between these. In one line of work, activity in speech motor areas rose with how loud participants experienced their inner speech, while activity in auditory perception areas fell [1]. That pattern fits a system that is simultaneously generating a motor plan and dampening the sensory expectation of hearing it. The motor component is not incidental; it tracks something about the subjective experience itself.
The mind's ear and the inner voice can come apart
The strongest argument against treating inner speech as just auditory imagery comes from people who report no conscious auditory imagery at all, a condition called anauralia. In a series of studies, these individuals showed intact phonological loop function across the classic working-memory effects, including phonological similarity, word length, and articulatory suppression [2]. In plain terms, the silent rehearsal system that holds words in mind worked normally even without any felt experience of hearing them.
That result suggests the inner voice is not simply the mind's ear pointed at words. The researcher proposed a "lexical ear" mechanism, a language-based route that supports verbal working memory independently of vivid auditory experience [2]. Large-scale online testing of over two thousand people found no effect of low auditory imagery vividness on verbal or musical working-memory spans [2]. So the felt sound of inner speech can be absent while the language work it seems to do carries on.
Motor areas mark inner speech, but the picture is not simple
Neuroimaging points to a specific motor region as a candidate marker for inner speech. The pre-supplementary motor area, a part of the brain involved in planning movements, showed activity linked to inner speech, and a single case of a person with absent inner speech showed significantly reduced activity there [2]. Earlier imaging work also implicated the supplementary motor area in tasks involving inner speech and auditory verbal imagery [3].
Work on speech motor imagery adds a further wrinkle: whether motor cortex is recruited appears to depend on the task, with some results suggesting that inner speech can proceed through routes that bypass motor cortex in certain conditions [4]. A cognitive neuroscience account tries to reconcile this by describing inner language as multisensory and motor, combining a "mind's ear" with inner phonation and articulation, and framing inner speech as a predictive control process rather than a single faculty [5]. On that view, the motor and auditory sides are partners in one system, and which one dominates may shift with the task.
So which description fits better?
The evidence favors a hybrid answer with a clear weighting. Inner speech behaves like a language process that normally includes motor components, and it can also carry auditory imagery, but the imagery is not required for the language function [2][1]. The motor activity scales with how loud inner speech feels [1], which suggests the motor side contributes to the experience rather than merely accompanying it, though that remains an interpretation rather than a settled mechanism.
The auditory imagery view is not wrong so much as incomplete. It describes one vivid layer of inner speech that some people have strongly and others barely have at all, while the underlying verbal machinery keeps running [2]. The motor view captures the silent-articulation flavor that imaging studies detect [1][3], but task-dependent findings mean motor cortex is not always in the loop [4]. The most defensible summary is that inner speech is a language process with motor components, one that often, but not always, comes wrapped in auditory imagery.
Sources used in this answer
Cortical activity increases in speech motor areas as a function of the subjective loudness of inner speech
Louder subjective inner speech was associated with increased activity in speech motor areas and decreased activity in auditory perception areas.
Mapping the Silent Mind: An Investigation of Auditory Imagery Vividness and Its Cognitive and Neural Correlates
Auditory imagery splits into a 'mind's ear' for external sounds and an 'inner voice,' and people lacking conscious auditory imagery still showed intact phonological loop function, with the pre-supplementary motor area proposed as a neural marker for inner speech.
Functional anatomy of inner speech and auditory verbal imagery
PET imaging of silent articulation and imagined speech implicated the supplementary motor area, differentiating motor and auditory imagery aspects of inner speech.
The cognitive and neural mechanisms involved in motor imagery of speech
Speech motor imagery research indicates motor cortex recruitment varies with task, informing whether inner speech necessarily involves motor components.
A cognitive neuroscience view of inner language
Inner language is described as multisensory and motor, combining a 'mind's ear' with inner phonation and articulation in a predictive control account.