What actually differs in the brain
The clearest contrast comes from a review of 83 studies on inner speech. When researchers ask participants to generate inner speech on cue, the left inferior frontal gyrus — a frontal region associated with producing speech — shows stronger activation. Spontaneous inner speech, the kind that arises without a prompt, instead engages Heschl's gyrus, an area on the auditory cortex involved in processing sound [1]. In plain terms, the prompted version looks more like the brain preparing to speak, while the spontaneous version looks more like the brain hearing something.
This is a difference in where activation is strongest, not proof that the two forms are entirely separate systems. The review frames these as distinct correlates of different forms of inner speech rather than as a complete account of how each works [1]. A separate decoding study compared elicited and self-generated inner speech and found that decoders trained on one condition could transfer to the other, which suggests the two share enough neural structure to be read by the same model even though their activation patterns are not identical [2].
Why a lab prompt may not capture everyday inner speech
Most experiments need inner speech to happen on schedule, so they elicit it: a participant is told to silently repeat a word or imagine a sentence. That makes the timing measurable, but it also imposes a task. The reviewed evidence indicates that this elicited form leans on frontal speech-production regions more than spontaneous inner speech does [1]. If everyday inner speech is closer to the spontaneous form, then a study built on elicited inner speech is measuring a somewhat different event.
Work on bilingual inner speech makes a related point about method. It notes that elicited inner speech may be more concrete and closer in nature to overt speech, whereas spontaneous inner speech is something else again [3]. That is a claim about the character of the experience, not a brain-imaging result, but it lines up with the neural contrast: the prompted version behaves more like outward speech.
How much does the lab setting distort the picture?
The gap is real but should not be overstated. The transferability of decoders across elicited and self-generated conditions suggests the two forms are not alien to each other [2]. What differs is emphasis — which regions are most engaged — rather than a wholesale switch to a different system. The review also notes large individual variation in how often people report spontaneous inner speech [2], so group averages may hide people whose inner lives are dominated by one form.
A scoping review of voice and speech as digital biomarkers adds a caution from a neighboring field. It observes that controlled elicitation dominates over naturalistic capture, and that strong predictive performance does not by itself imply ecological validity or clinical deployability, since biomarkers still need to be shown to hold up in spontaneous, multispeaker settings [4]. The same logic applies to inner speech: a clean lab signal is not automatically a faithful picture of the spontaneous case.
What to take away if you are reading inner-speech research
When a study reports brain activity during inner speech, the first question to ask is how that inner speech was produced. If participants were cued, the findings most directly describe elicited inner speech and its frontal, speech-production weighting [1]. If the study captured inner speech as it arose naturally, the findings are more likely to reflect auditory involvement [1]. Neither is wrong; they answer slightly different questions.
The practical upshot is that lab findings on inner speech are informative but not a complete map of the everyday phenomenon. The reviewed literature supports treating elicited and spontaneous inner speech as related but distinguishable, and it flags emotional aspects, developmental trajectory, and clinical relevance as areas still needing work [1]. For now, the honest summary is that the two forms differ in their neural emphasis, and that this difference is a reason to read prompted studies as evidence about prompted inner speech first.
Sources used in this answer
The multifaceted nature of inner speech: Phenomenology, neural correlates, and implications for aphasia and psychopathology
A review of 83 studies reports that task-elicited inner speech activates the left inferior frontal gyrus more strongly, while spontaneous inner speech engages Heschl's gyrus, suggesting auditory involvement.
A note on decoding elicited and self-generated inner speech
Decoders of inner speech transfer across elicited and self-generated conditions, and the paper notes large individual variation in how often people experience spontaneous inner speech.
The Little Voice in the Bilingual's Head: Inner Speech–What Language Does It Speak? and How It Is Affected by Emotions
In bilinguals, elicited inner speech may be more concrete and closer to overt speech, while spontaneous inner speech differs, and ecological momentary assessment is described as reducing recall bias.
Voice and speech as digital biomarkers of mental health: a scoping review of clinical interpretability, data collection and translational readiness
A scoping review finds controlled elicitation predominates over naturalistic capture in speech biomarker research, and that predictive performance does not by itself establish ecological validity or clinical deployability.