Personal music and photographs as cues for autobiographical re-experiencing

01 Scope and theoretical basis

Autobiographical memory concerns experiences from a person's own life. Janata, Tomic and Rakowski (2007) documented both general and specific autobiographical content in responses to music. Jakubowski and Francini (2023) found that familiarity affected how readily memories were retrieved, while the emotional expression of music affected other properties of the memories. These studies informed the choice to measure several aspects of the response separately.

VISIONSeed compared personally associated songs and photographs as they existed in the participants' archives. Familiarity, prior exposure and the information directly available in each medium differed. The study therefore concerns these personal cue packages rather than isolating hearing from vision under otherwise identical conditions.

The VISION study narrowed the question to responses after personal cues, and ARC concerned the subsequent interaction. No neural or physiological measurements were collected. The relevant earlier periods had not been measured while they were occurring.

The reminiscence bump concerns the distribution of autobiographical memories across the lifespan. It was part of the project's background research, but this five-person study did not test that distribution.

02 Research question and hypotheses

Does first-cue modality change reported autobiographical re-experiencing when the cues are drawn from a person's own music and photographic archive?

For the conditional allocation analysis, the sharp null states that assigning a prepared context to music rather than photography would not change its response. The alternative permits a difference in either direction. The test is two-sided.

Memory form, affect, broad chapter association and past-self connection were secondary outcomes. They were not combined into a return percentage or composite success score.

03 Participants and units

Six young adults took part. P01, P02, P04, P05 and P06 completed six encounters each; P06 was the investigator-participant. P03 left partway through giving their cues, so their session is incomplete and is excluded from all encounter denominators.

The participant is the unit used for the group contrast. Thirty encounters supply repeated observations within those five participants, not thirty independent replications. The sample was selected for feasible personal archives and cannot represent the population of music listeners.

04 Cue preparation and allocation

Each participant supplied two distinguishable chapters and three contexts within each chapter. Every context had an associated song and photograph. Historical listening or image metadata and participant-supplied meaning were kept conceptually distinct.

The intended constrained allocation gave each participant three music and three photograph first encounters, with both modalities represented in each chapter. The delivered schedules appear in the allocation table. Only the first cue for each context enters the analysis; this was not a same-context crossover.

Songs and photographs were authentic personal media, not sensory-equated laboratory stimuli. Familiarity averaged approximately 5/6 for music and 4.2/6 for photographs. Such differences remain possible explanations for the observed contrast.

05 Encounter procedure and amendment

Preparation, encounter responses and post-session evaluation were collected using three Google Forms created through Google Apps Script. Cue allocation and presentation were managed separately by the operator. The prototype was subsequently used for formative interaction tasks; the interface shown in the portfolio is a revised demonstration.

The sequence was readiness, pre-cue pleasantness and activation, a thirty-second music excerpt or photograph, uninterrupted report, then structured measures. The operator confirmed that song playback stopped at thirty seconds and all photograph metadata was concealed. Some written excerpt endpoints were inconsistent, but the later delivery confirmation resolves that procedural question. Complete instrument-level timing logs were not supplied.

The free-report prompt was: “Describe whatever came to mind or changed during the cue. Include uncertainty. You don’t need to make it fit any particular period.” Detailed questions followed that account. Post-encounter affect therefore reflects the cue and narration together.

The complementary-cue block was removed prospectively after session burden became apparent. The final analysis concerns first-cue responses and cannot estimate the incremental benefit of presenting both modalities or an optimal sequence. The protocol remained named VISIONSeed v4 after this amendment.

06 Measures

“During that cue, how strongly did you feel you were experiencing a moment from your own past again?”

Reported re-experiencing was the primary measure. Its endpoints were 0, not at all; 6, extremely strongly. Secondary records included participant-selected memory form, broad chapter association, pleasantness and activation, past-self connection, material not directly supplied by the cue and readiness to continue. These are distinct observations with distinct interpretations.

An access button was optional. No latency analysis is reported because a complete set of access-button timings is unavailable. Unrecorded times were not estimated from participants’ descriptions.

07 Primary calculation

Within a chapter, one cue type appeared twice and the other once. The calculation first compares the two cue types within that chapter, then gives the participant's two chapters equal weight. The group result averages the five participant differences.

For participant p and chapter c, Dpc is the mean music rating minus the mean photograph rating. Dp is the mean of that participant's two chapter contrasts. The group estimate is the mean of the five Dp values.

The participant contrasts 0, .5, 1.5, .75 and 1 sum to 3.75 and average .75. P01’s zero contains opposing chapter effects of +2 and −2. The raw music total is 68 over 15 observations and the photograph total is 57 over 15, giving means of 4.5333 and 3.8. Their pooled difference, .7333, differs from .75 because the primary calculation gives equal weight to each chapter before averaging participants.

Uncertainty and assignment assumptions

08 Conditional allocation-model sensitivity

The intended mechanism permits 18 assignments per participant. Across five participants, 18⁵ yields 1,889,568 combinations. The reported number with an absolute statistic at least as large as .75 is 139,782; their ratio is .0739756.

The executable generator and complete assignment provenance were not preserved sufficiently to establish that every actual schedule was sampled from this mechanism. The result is therefore conditional on the intended allocation model; it is not a verified exact randomization p-value.

Removing P06 gives a mean of .6875 and reported conditional p = .167. A participant t-model sensitivity in the source gives a full-sample p near .040. This calculation relies on different assumptions. Both remain visible, and neither resolves the substantial uncertainty of five participant effects.

The p-value describes how extreme this result is under the stated no-effect hypothesis and intended assignment procedure. It is not the probability that the hypothesis is true. The calculation remains conditional because the full executable assignment procedure was not preserved.

In plain words

If cue type made no difference, every situation would have produced the same rating whichever cue it received. Under that assumption I can reshuffle which situations got songs, within the rules the allocation followed, and recompute the result each time. There are 1,889,568 possible allocations. 139,782 of them produce a difference at least as large as the one observed, in either direction.

This p-value describes how extreme the result is under the stated no-effect hypothesis and the intended allocation procedure. It is not the probability that the hypothesis is true, and because the exact assignment code was not preserved, it is conditional on that intended procedure. Five people cannot settle a general difference between music and photographs.

09 Secondary outcomes

Photographs received particular-event or scene classifications in 10/15 encounters and music in 6/15. Both received broad prepared-chapter associations in 14/15. A chapter mismatch with re-experiencing 5/6 shows why intensity alone does not identify destination. These classifications describe what participants reported; they do not establish historical accuracy.

Current activation increased by a descriptive mean of 38.67 points for music and 30.67 for photographs; pleasantness changed by +9 and −1 respectively. Past-self connection averaged 4.53 for music and 3.73 for photographs on the 0–6 scale, a raw secondary difference of +.80. These cue-and-report changes lack a no-cue comparison.

Positive convergence under all four related affect checks occurred in 13/15 photograph encounters and 9/15 music encounters. These calculations reuse the same observations under related choices. They are sensitivity checks, not independent replications, and describe a different outcome from reported re-experiencing.

10 Affect distance and its ambiguity

The worked figure is a mathematical illustration. Chapter A = (80, 60), Chapter B = (60, 60), before = (70, 30), and after = (70, 75). Both axes run from 0 to 100. Normalised absolute distance is the sum of the two axis differences divided by 200.

In the illustration, the initial point differs from either reference by 10 pleasantness points and 30 activation points. Dividing their sum by 200 gives 0.200. Afterward, the differences are 10 and 15, giving 0.125. Both references become equally close, so the calculation does not distinguish which period was recalled.

Distance to either chapter reference falls from .200 to .125, giving convergence of +.075 for both. Relative preference stays zero. Becoming closer to a remembering reference cannot, by itself, identify an autobiographical destination. The references concern how a chapter feels when remembered now; they are not recordings made during the original period.

11 Free reports and record handling

P01 E02’s report was spoken in order but absent from the original form; it survives in an operator note. The quotation below preserves that provenance. P04 B3’s stored photo inventory and recalled context disagree. The original record and later interpretation are shown separately; an editable product association does not overwrite the study record.

Narrative material was not independently coded by blinded human raters, so no validated thematic counts are reported. The anonymous quotations retain the participants’ wording. The shorter P04 quotation is identified as a contiguous excerpt.

12 Formative prototype tasks

The tasks examined intention selection, explanation inspection, association correction, unexpected continuation and future resurfacing. They were administered under varying conditions. These formative prototype observations form a separate record from the externally collected experiment.

The table motivates making correction and resurfacing clearer. It does not validate the revised interface or establish a causal usability improvement. No overall success percentage is reported.

When asked about an unexpected association, three participants preferred being asked whether to follow it and two preferred returning to the selected context. None preferred automatic following. These responses informed the proposed continuation choices. They do not establish that the revised interface has improved outcomes.

Resulting design implications

13 Product implications

Atlas separates media provenance from personal association. ARC offers choices after the person's account rather than optimising a presumed internal state. The revised design makes cue replacement, explanation, uncertainty, correction, stop and future resurfacing explicit.

An editable product association cannot overwrite the frozen experimental record. Stopping the current encounter cannot silently become a permanent suppression rule. Future restrictions identify their scope and remain reversible. A design proposal can make these choices precise while remaining open to subsequent evaluation.

14 What this can claim

In this group, music drew stronger reported re-experiencing, and photographs more often returned a specific scene.

15 Limitations and further evaluation

The study is small, uses a convenience sample with an investigator-participant, and compares stimuli that differ in familiarity, previous exposure and directly supplied information. There is no no-cue deliberate-recall condition. Equipment and volume were not fully standardised; complete instrument-level timing logs are unavailable. Affect follows cue and narration, while narrative content lacks independent blinded coding.

A replication should preserve an executable allocation procedure, match familiarity more closely, include deliberate recall without a cue and independently code narratives. Revised interface evaluation should separate ordinary use from experiment administration and report help, failures, burden and unwanted resurfacing without collapsing them into one score.

16 Future Seed

Future Seed would record a present period with repeated self-report and selected media, reserve cues and compare later responses with a genuine earlier reference. Neural or physiological measurement would require a separately justified protocol and cannot retroactively supply missing historical ground truth.

The completed study supports a more precise description of present-day remembering. The prospective proposal asks which recorded components can recur after a delay. They remain different evidence claims.

The full study design

1. Phase 1 records reference states without music.
2. Phase 2 pairs new, meaningless sound keys with those states.
3. Phase 3 freezes a decoder trained only on the silent references.
4. Phase 4 crosses the person’s chosen target with a concealed key:

The same chosen target is crossed with one of three concealed keys: matched, rival (similar feeling, different state) or unpaired. Each is followed by a silent window, where the margin is read. The hypothesis is that the margin is highest after the matched key.

Because the intention is identical in every arm, any difference belongs to the sound. A positive result would need all of these at once:

• the margin rises more for matched keys than for rival and unpaired ones;
• it favours the target over every rival state;
• it appears in the silent window after the sound;
• it generalises across keys, episodes and days;
• it survives controls for acoustics, pleasantness and activation;
• it converges across at least two measurement layers;
• it produces no increase in false detail or distress.

Terms

Margin

How similar the present moment is to the state being aimed for, minus its similarity to the closest rival state.

Decoder

The model that recognises each state, trained only on the references recorded without music and fixed before any key is played.

Silent window

The silence after the sound, when the margin is read.

Measurement layers

The kinds of measurement a state can be read from: brain, body and behaviour.

17 Sources and records

Primary records: Start-operator-mode (2), including delivered schedules, verbatim reports and amendments; Experimental-design-greeting (5), completed numerical analysis and audit; Vision-experiment-context (3), prototype and interim evidence; VISION-(porto), historical conceptual development.

The encounter tables reproduce numerical records and analyses documented in the study manuscript and operator transcripts. Original form exports are not included in this public reader.

Janata, P., Tomic, S. T., and Rakowski, S. K. (2007). Characterization of music-evoked autobiographical memories. Memory, 15(8), 845–860. doi:10.1080/09658210701734593. Published reports distinguish general and specific autobiographical content.

Janata, P. (2009). The Neural Architecture of Music-Evoked Autobiographical Memories. Cerebral Cortex, 19(11), 2579–2594. doi:10.1093/cercor/bhp008. This separate neural study does not supply neural evidence for VISIONSeed.

Jakubowski, K., and Francini, E. (2023; online 2022). Differential effects of familiarity and emotional expression of musical cues on autobiographical memory properties. Quarterly Journal of Experimental Psychology, 76(9), 2001–2016. doi:10.1177/17470218221129793. Familiarity and emotional properties motivate retaining distinct outcomes and qualifying the modality comparison.