Why audio-first review can work, and what we still have to prove.

In 2012, researchers at Microsoft Research Asia and three universities built MemReflex, an adaptive flashcard system for mobile microlearning.3 Twelve participants learned factual material (animal lifespans) while walking, using either audio-only or text-only cued-recall flashcards.

Limits. The authors report no significant difference between the two. The study was small, each task lasted about five minutes, and the memory test came only two minutes later, so it doesn’t establish long-term equivalence.

Another finding may matter as much: when the walking task added visual demands, accuracy on the repetition test dropped more for audio than for text. The researchers argued for switching between modalities as the learner’s context changes. That matches FlashPod’s principle: audio-first, not audio-only. Audio learners also walked about 15% faster than text learners.

Source: Edge, Fitchett, Whitney & Landay, MobileHCI 2012 [3].

A 2022 meta-analysis combined 46 studies and 4,687 participants comparing reading and listening comprehension.4 Overall comprehension wasn’t reliably different between the two formats, and literal comprehension was essentially equivalent. Reading had an advantage when inferential comprehension was tested and when reading was self-paced, though the self-paced effect was small.

Limits. The comparison is of reading and listening comprehension, not of flashcard review.

FlashPod keeps Visual review next to audio because the best modality depends on both the card and the context.

Source: Clinton-Lisell, Review of Educational Research, 2022 [4].

FlashPod doesn’t simply play answers. A 2017 meta-analysis of practice-testing research found that practice tests improved learning compared with restudying and other non-testing conditions.5 A major quantitative review of spacing reached a complementary conclusion for verbal recall tasks: when learning episodes occur matters for later retention.6

A 2013 foreign-vocabulary study compared retrieval practice with imitation, repeating after hearing the answer, and favored retrieval, with tests given right away and after a two-day delay.7 That is the choice an audio review makes on every card: ask first, or just play the answer.

Together these support the loop underneath FlashPod: retrieve, get feedback, bring the material back later.

Limits. These findings aren’t specifically about audio delivery. They support a learning loop that FlashPod applies through headphones or on a screen.

Sources: Adesope, Trevisan & Sundararajan, 2017 [5]; Cepeda et al., 2006 [6]; Kang, Gollan & Pashler, 2013 [7].

A randomized trial of 137 emergency-medicine trainees compared two versions of the same educational podcast; one included five retrieval questions.8 Within 48 hours there was no significant overall difference, although the questioned items themselves already scored higher. Two to three weeks later, the group that received questions scored 5.6 percentage points higher overall (the confidence interval was wide) and 10.1 points higher on the material that had actually been questioned.

Limits. This wasn’t a FlashPod study. It tested a podcast with embedded questions, not scheduled flashcards. It does show that adding retrieval questions to audio can change later retention, beyond making listening feel more interactive.

Source: Weinstock et al., Annals of Emergency Medicine, 2020 [8].

FlashPod gives you time to retrieve the answer mentally before you hear it. A 2025 meta-analysis of 18 studies and 2,560 participants found that silent retrieval produced a small but significant learning benefit.9 Producing the answer overtly (writing, typing or speaking) was somewhat more effective, but silent retrieval still helped.

Limits. Silent and overt retrieval aren’t identical. The practical point is narrower: trying to remember in your head still counts as retrieval practice.

Source: Yu et al., Educational Psychology Review, 2025 [9].

FlashPod doesn’t claim that walking improves memory or that multitasking is free. The useful question is whether light or routine movement necessarily prevents useful learning. In a 2024 randomized crossover trial, 96 emergency-medicine residents listened to a 30-minute educational podcast seated and another during 30 minutes of continuous aerobic exercise.10

Recall after podcast listening (initial test within 30 minutes; delayed test at 30 days)

Test	Exercising	Seated
Immediate recall	74.4%	76.3%
30-day recall	52.3%	52.5%

Neither difference was statistically significant.

Limits. The study used passive podcast listening, not flashcard retrieval, so it can’t show that FlashPod review while moving equals seated study. Combined with the MemReflex experiment it supports a narrower conclusion: useful verbal learning can occur during some forms of movement.

Context still matters. Use FlashPod only when it’s safe to do so; some situations demand your full attention, and FlashPod isn’t meant for those. It aims to make the screen optional when the situation allows, not to make every situation right for studying.

Source: Gottlieb et al., Academic Medicine, 2024 [10].

Written information stays put for you to reread; speech doesn’t. That makes length and segmentation especially important for audio. Research on the transient-information effect suggests that long spoken passages can overload working memory,11 and a meta-analysis of 56 investigations found that dividing multimedia instruction into meaningful segments improved retention and transfer and reduced cognitive load.12

Limits. This research was about multimedia instruction in general, not flashcards or audio specifically, so it would be too strong to call flashcards inherently ideal for audio. The design implication still holds: short question-and-answer units, deliberate pauses, replay and learner control fit spoken information better than long uninterrupted passages.

Sources: Leahy & Sweller, 2011 [11]; Rey et al., Educational Psychology Review, 2019 [12].

FlashPod doesn’t claim audio makes learning faster; speech can be slower than reading. The opportunity is that scheduled review becomes possible in moments when continuously watching a flashcard app would be inconvenient.

In one two-week field study, 20 people used WaitChatter, a Google Chat extension that showed short vocabulary quizzes while they waited for instant-message replies. In a post-study quiz they could recall an average of 57 new words learned during those brief waiting moments.13

Limits. That study was visual, not audio. It had no control group and measured recall on a quiz afterward, not long-term retention, and it doesn’t prove FlashPod will increase study time. It shows something narrower: small retrieval opportunities work when they fit the day.

Source: Cai et al., CHI 2015 [13].