Last lesson distinguished a general description of your thinking from a report about a particular moment. This lesson takes the next step: interpreting a series of momentary reports. Repeated sampling can reveal a pattern that a questionnaire might miss, but even a record with no reported inner speech has boundaries. Allow about 35–40 minutes for the selected passages and the worked example.
What a sampling record actually contains
Imagine an alert sounding at an unpredictable time. The report concerns what was experienced just before the alert, not what the person was thinking earlier in the day or what they think about while answering. That narrow target is a strength: it reduces the need to reconstruct an entire day from memory. It is also the first limit on the conclusion. An alert samples a moment; it does not monitor the interval between alerts.
Watch the distinction between experience at the signal and an account reconstructed afterward.
What is your internal world really like? (Russell Hurlburt) | Inner Cosmos with David Eagleman
In “What is your internal world really like?” on Inner Cosmos with David Eagleman, Russell Hurlburt explains why the timing of a report matters. Watch for the difference between describing a thought caught by the signal and describing a habitual pattern.
Watch the sampled instant for his definition of what must be present at the beep. Then watch later descriptions for examples of reports drifting toward reactions to the beep or statements about what usually happens.
The phrase “I was thinking about lunch” can be a useful note, but it does not yet establish that words were present. It names a topic. A more detailed account might distinguish silently saying “I should eat,” seeing food in imagination, or having a thought about eating without experienced words or an image. Classification matters particularly when a record appears to contain no inner speech: some entries may be clear reports of its absence at that moment, while others may simply lack enough detail to tell.
DES addresses this problem by returning to sampled moments in interviews, rather than treating the first label as final. Read the account of how those descriptions are developed.
Pristine Inner Experience and Descriptive Experience Sampling: Implications for Psychology
This article’s “Descriptive Experience Sampling” section explains how repeated interviews clarify reports without requiring every moment to fit a tidy category. It supplies a useful standard for judging how much confidence to place in a recorded “no inner speech.”
In “Descriptive Experience Sampling,” start with the paragraph beginning “During the expositional interview.” Read the interview account, ending with the paragraph about written descriptions. Notice that uncertainty can remain in the final description rather than being converted into a definite absence.
An interviewed DES record and a brief app record therefore need not carry equal detail. Neither is direct observation of another person’s experience. Both depend on a report, but one offers more opportunity to examine what the reporter meant.
A record with zero reported instances
Consider a fictional 12-day study. Each day, an app schedules one alert randomly within a morning window and one within an afternoon window. A participant responds to 16 of the 24 alerts and reports no inner speech at any of the 16 sampled moments.
| Alert window | Alerts sent | Reports received | Reports of inner speech |
|---|---|---|---|
| Morning | 12 | 10 | 0 |
| Afternoon | 12 | 6 | 0 |
| Total | 24 | 16 | 0 |
The direct result is zero reports of inner speech in 16 answered samples. That is informative: repeatedly asking about specific moments did not produce an affirmative report in this record. If the reports were accurate and the classification was clear, they give evidence that inner speech was absent at those particular moments.
The denominator matters. The participant answered 16 alerts, not 24. Writing “zero out of 24” would quietly treat eight missing reports as eight reports of absence. Likewise, “zero percent of waking thought involved inner speech” would replace a set of sampled moments with a claim about continuous experience. The record supports neither substitution.
The uneven response pattern matters too. Six of the eight missed alerts were in the afternoon. If responding was harder during particular activities, those missed moments might differ from the answered ones. The alerts also occurred only within the study’s chosen windows. Evening conversations, deliberate rehearsal, and other unsampled contexts remain outside this record.
For a real app study, read how its response instructions and alert windows restricted what could be observed, then compare its reported completion rate with the fictional record.
This study provides a concrete example of repeated alerts in daily life. The methodological details make it easier to see why “random alerts” does not mean “every context had an equal chance of being recorded.”
In “Materials and Methods,” under “Procedure,” read the alert procedure. Focus on the instruction to respond when safe and on the limited, participant-selected windows. Then, in “Results,” under “Overall Characteristics of Inner Speech,” read the response summary. Note that not every alert yielded a response; do not treat a missed alert as a negative report.
That app study asked about particular characteristics of inner speech at alerts. It was not designed to turn an unanswered alert—or a low rating for one characteristic—into proof that no form of inner speech occurred.
Why zero is evidence, but not proof of “never”
Even if every one of the fictional participant’s 16 reports were immediate, unambiguous, and accurate, a finite sample can miss an experience that occurs occasionally. To make the point concrete, suppose—for illustration only—that inner speech is present at 10% of eligible moments and that 16 moments are sampled independently from that same set of moments. The chance of sampling no instances would be:
In that simplified model, zero sightings are quite possible despite a nonzero rate. This calculation does not estimate the participant’s actual rate, nor does it assign a 19% probability to any claim about them. Real reports may cluster by activity or time of day; responses can be missed; and recognizing inner speech is not an error-free measurement. The calculation shows only why “not seen in this sample” cannot logically become “never occurs.”
There are also several distinct claims that a zero record cannot settle:
- Other times: Inner speech might occur between alerts or outside the permitted windows.
- Other contexts: It might occur during a task scarcely represented in the record, such as memorizing words or preparing what to say.
- Deliberate use: A person might be able to produce silent words on request even if none were reported spontaneously at sampled moments.
- Uncertain classification: A brief entry such as “thinking about tomorrow” may not reveal whether language was experienced at the instant sampled.
These possibilities do not require dismissing the zero record. The disciplined conclusion is narrower: inner speech was not reported at the answered sampled moments, under this study’s reporting procedure. The strength of any further estimate depends on coverage, response patterns, and how carefully the reports distinguished words from the topics of thoughts.
Keep group comparisons separate from individual records
The chart below illustrates why the method of asking is part of the finding. Across three studies, general questionnaire estimates and DES reports differ for several kinds of inner experience, including inner speaking.

This is a group comparison between methods, not a continuous trace of anyone’s mind. It does not show that every questionnaire answer is wrong or that DES detects every occurrence. A general estimate and a report tied to sampled moments put different demands on memory and description. Equally, a group-level DES frequency cannot tell you whether a particular person with zero affirmative reports would speak inwardly during a different week or task.
The distinction becomes especially important when assessing claims that someone has “no inner voice.” A general questionnaire may convey how they characterize themselves. Repeated sampling can test what they report across selected moments. Neither, on its own, establishes the absence of inner speech in every context.
Conclusion
A repeated-sampling record is best read from its actual observations outward: which alerts received responses, what was reported at those moments, and which times or situations were never captured. Zero affirmative reports can be meaningful evidence of infrequent reported inner speech within the sampled record. They are not proof of permanent absence, and missed or ambiguous entries must not be silently counted as negatives.
The next lesson examines a different kind of evidence: whether disrupting speech-related processing changes task performance. That experiment can reveal a contribution of verbal processing without directly showing what someone consciously experienced.