A minute can be measured accurately while its experience remains difficult to describe. Bergson's account of duration asks what is lost when a developing experience is represented as a row of separate, interchangeable units.
Duration and its symbols
In chapter II of Time and Free Will, Bergson distinguishes quantitative multiplicity from the interpenetration of conscious states. His discussion of a melody presents successive notes as experienced through their relation to what has already been heard. Isolating each note from the rest changes the experience being described.
He criticizes the representation of lived succession as though its moments were externally juxtaposed positions in a homogeneous space. His account of duration concerns qualitative continuity, while a clock provides a symbolic quantitative comparison. The argument should not be reduced to the observation that a boring hour feels long.
The text is a philosophical proposal about experience. It is not a formula for clock drift, an experimental refutation of time measurement or a derivation of a second physical temporal coordinate.
The same elements in a different order
Consider an original example involving three tones, each lasting one second. One sequence rises from low to middle to high; another falls from high to middle to low. Their total durations and tone counts are identical, but their order differs. A summary that stores only counts cannot distinguish them.
A richer record can preserve order. Another can preserve loudness, timbre and pauses. Adding these details improves the account of the acoustic signal, but it does not by itself settle whether the resulting description exhausts the listener's experience.
This separates two criticisms. A record may be inadequate because it discarded measurable structure. A philosophical critic may additionally argue that even a complete physical description leaves something about experience unexplained. The first criticism can be demonstrated by a counterexample; the second requires its own argument.
A model of retention is not lived duration
Take the three numerical inputs 1, 2 and 3 and a simple memory update m(n + 1) = m(n)/2 + a(n + 1), with m(0) = 0. In ascending order, the successive memory values are 1, 2.5 and 4.25. In descending order, they are 3, 3.5 and 2.75.
The same sum of inputs, six, therefore produces different final states. This stipulated weighted memory retains something about order that an unweighted total loses. It makes an informational distinction visible through a short calculation.
It would be a mistake to rename the memory variable lived duration and declare the philosophical issue solved. The update has no supplied account of awareness, musical meaning or qualitative experience. Its value lies in the exact, narrower result it establishes.
Measurement and description can cooperate
A careful investigation can combine an accurately timed performance with a listener's account of how its phrase develops. The timing identifies a reproducible physical condition; the report supplies evidence about experience. Neither should be silently substituted for the other.
Different listeners may report different experiences of the same recording. That raises questions about familiarity, attention, expectation and method. It does not imply that the recording had a different physical duration for each listener. Equally, a shared duration does not establish identical experiences.
The distinction makes interdisciplinary research more precise. Specify which features of the stimulus are controlled, which reports are compared and which interpretation of those reports is being tested. The result can then constrain a particular account without claiming to settle every philosophical question about time.
Renewal with retained history
The SMM/DTT book treats persistence through relations between successive records. This opens a useful comparison with questions about continuity: renewing a carrier need not erase its correlations. The particular form of retention still has to be specified.
The book's future-test criterion asks which histories remain distinguishable by later observations. The two input orders above cannot be merged if the final memory can be measured. That is an operational result, not an identification of Bergson's duration with inner time.
A productive comparison preserves this difference while asking what a generative model can contribute to the study of temporal continuity. The essay on Augustine's memory and expectation provides another distinct starting point.
Sources and further reading
- Henri Bergson, Time and Free Will, chapter II, especially the discussion of duration and melody around printed pages 100–106; F. L. Pogson translation, 1913 edition.
- Mohamed Haj Yousef, Renewal, persistence, and a continuous limit and When are two histories the same?. The numerical memory model is a teaching construction.