The More Accessible Information Is, the Less Employees Remember

Esther Kang · article · 2026-09-22 Reader 🖥️

Cognitive Surrender Desirable Difficulty & Productive Struggle Learning Culture & Slack Transactive Memory (Network Memory vs. Content Memory)

Because greater connectivity to people, platforms, and AI assistants makes employees measurably better at knowing who or what holds information and measurably worse at retaining the information itself -- even among their highest-capacity people -- leaders must stop treating findability as a proxy for learning and instead deliberately engineer work (explaining, applying, debating, teaching) that forces genuine internalization.

practical · mixed — Opens as an expository research report (summarizing the author's own five-experiment study) but is structured throughout to argue for managerial action -- the research exists to license three explicit 'What Managers Should Know' prescriptions. The theoretical claim (connectivity trades content memory for network memory) is instrumental to the practical thesis (leaders must design work to force internalization).

Structure

Part	Role
Opening: the connectivity investment and its unexamined assumption ('greater connectivity leads to better learning')	context
Description of the author's own research program: 5 experiments, 1,000+ adults aged 18-77, simulated organizational/online networks, measuring content memory vs. network memory	evidence
'Why We Remember People Instead of Information' -- the new-team-member anecdote plus prior psychology on distributed/social knowledge, framing the finding as an efficient, rational strategy rather than a deficit	explanation / mechanism
The headline results: ~40% less content recall / ~65% better network recall after forming connections; the effect is strongest, not weakest, among high-working-memory-capacity participants	evidence
Three 'What Managers Should Know' sections (access is not learning; best employees may offload most; turn social networks into learning networks)	conclusion / practical prescription
Closing section on the AI workplace: extrapolates the tradeoff forward as retrieval cost keeps falling toward zero	extension / open question

Key terms

the connectivity tradeoff
The author's own coined term for the shift she measured: as people become more socially/digitally connected, they become less likely to learn the information their network shares and more likely to remember only where it resides -- treating the network itself as an external memory system.
content memory Transactive Memory (Network Memory vs. Content Memory)
Remembering the substantive information itself -- 'who knows what' as a fact learned, not as a social pointer.
network memory Transactive Memory (Network Memory vs. Content Memory)
Remembering relationships between people -- 'who knows whom' / who is connected to whom in the social graph -- as distinct from remembering the content those people hold.
cognitive offloading Cognitive Surrender
Used in its established psychological sense (cited via Risko & Gilbert) -- devoting less internal effort to encoding information because it can be externally retrieved -- applied here specifically to offloading onto one's social network rather than onto notes, devices, or AI.

Propositions

explicitGreater connectivity produces a cognitive tradeoff: as people become more connected to others, they become less likely to learn the information those people share and more likely to remember where that information resides and who holds it.conf: high
extendsCognitive Surrender · p5 — Extends the AI-writing finding (reduced memory of one's own AI-assisted output) to a purely social, non-AI mechanism -- showing connectivity itself, not just AI use, drives offloading onto an external memory system.
explicitAcross five experiments with more than 1,000 adults aged 18-77, participants who formed social connections before learning new material remembered about 40% less of the content but were about 65% better at recalling who was connected to whom, and the pattern held consistently across studies.conf: high
extendsAI coding creates two kinds of debt. You're only measuring one · p8 — Generalizes the coding-specific observation that AI accelerates loss of the 'mental map of who knows what' into a broader, experimentally-demonstrated connectivity tradeoff between content memory and network memory that holds even without AI in the loop.
supportsDet skal være vanskelig · p6 — Supplies direct experimental evidence for Bjork's desirable-difficulties mechanism: lowering the effort of retrieval (via social connectivity, not AI specifically) measurably lowers content retention, exactly as 'the effort is the mechanism' predicts.
explicitIndividuals with higher working-memory capacity showed the strongest version of this shift: more than a 50% reduction in content recall alongside more than doubling their accuracy in remembering social connections.conf: high
explicitAccess to information does not automatically translate into learning; the easier it becomes to retrieve information from other people (or systems), the less likely employees are to retain it themselves.conf: high
explicitThe shift toward network memory is a rational, strategic reallocation of cognitive effort rather than a failure of attention or an overload response -- participants were not forgetting because they were overwhelmed, they were redirecting effort toward the social map.conf: high
explicitLeaders should distinguish between information employees can simply look up and knowledge employees need to internalize for sound judgment and creative problem-solving, and should evaluate systems on whether employees still learn, not just whether they can find information.conf: high
explicitRequiring employees to explain ideas in their own words, discuss new knowledge with colleagues, apply it to real problems, or teach it to others produces deeper learning than simply making information available, and organizations should build these activities into work deliberately.conf: high
implicitHigh performers are not more resistant to offloading than average employees -- they may be the fastest to recognize that memorizing is no longer the most efficient strategy, so 'this person always knows the answer' can mean 'knows where to look' rather than 'has internalized it,' undermining a common managerial assumption that top performers retain more.conf: high
implicitBecause the tradeoff is a rational response to falling retrieval cost rather than a fixable behavioral lapse, investing further in collaboration platforms and AI assistants -- without any structural change to how work is organized -- will continue to erode employees' internalized knowledge base over time; the effect compounds with connectivity rather than self-correcting.conf: medium
implicitThe author treats 'knowing where to find information' and 'genuinely possessing/understanding it' as a clean binary state an employee is in, rather than a continuum or a state that can degrade gradually or be partially true -- 'access to knowledge is not the same as possessing it' is asserted as a sharp line, not a matter of degree.conf: medium

Arguments

Leaders should evaluate knowledge and collaboration systems not simply by whether employees can find information, but by whether employees still learn it -- and should build active-processing requirements into work rather than only investing in better retrieval infrastructure. strength: medium

Explicit premises:

Digital workplaces have dramatically reduced the cost of retrieving knowledge (search, collaboration platforms, AI assistants).
The author's five-experiment research shows connectivity shifts effort from content memory to network memory.
Workplace activities like innovation, strategic thinking, and decision-making depend on integrating knowledge from multiple sources, not merely retrieving it on demand.
Requiring employees to explain, discuss, and apply information produces deeper learning than access alone.

Implicit premises:

Synthesis and integration of knowledge cannot happen purely through real-time lookup -- it requires some baseline of internalized content that can be combined and reasoned over.
The cost of retrieval-dependence (slower synthesis, weaker independent judgment) is not visible in, or measured by, standard organizational metrics, so it accumulates without being noticed or corrected.
An organization that optimizes purely for findability is implicitly trading a harder-to-measure capability (internalized judgment) for an easier-to-measure one (speed of access), without anyone deciding to make that trade.

Inferences:

If the tradeoff is a rational adaptation to a low-retrieval-cost environment, exhorting employees to 'try harder to remember' without changing the environment is unlikely to work -- the fix has to change what's effortful by design (as the article's own prescriptions do), not appeal to individual willpower.
The tradeoff should get worse, not better, as AI assistants keep lowering the marginal cost of retrieval toward zero -- so the mitigation the author proposes needs to be a permanent structural feature of how work is organized, not a one-time initiative.

Managers should not assume that high-performing or high-working-memory-capacity employees automatically retain critical knowledge simply because they can access it easily.