Latest / Mind Cast / Mental Model Inertia | Why Tech Rollouts Fail (And How to Fix Them)

Transcript

Mental Model Inertia | Why Tech Rollouts Fail (And How to Fix Them) · Mind Cast

• 0:00
  SPEAKER_00: April 26, 1994.
• 0:03
  SPEAKER_00: Nagoya, Japan.
• 0:05
  SPEAKER_00: China Airlines Flight 140 is on final approach.
• 0:09
  SPEAKER_00: The weather is clear.
• 0:11
  SPEAKER_00: The aircraft is mechanically perfect.
• 0:14
  SPEAKER_00: There is no malfunction.
• 0:16
  SPEAKER_00: And yet, 264 people are about to die.
• 0:21
  SPEAKER_00: Not because the plane broke, not because the pilots were careless, but because one pilot's brain, expertly trained, thousands of hours forged, was still operating inside a cockpit that no longer existed.
• 0:36
  SPEAKER_00: In the gap between his old mental map and the new machine, a catastrophe was born.
• 0:42
  SPEAKER_00: That gap has a name.
• 0:44
  SPEAKER_00: And by the end of this episode, you'll understand what it is, where it lives, and why it doesn't just affect pilots, it affects doctors, engineers, soldiers, and anyone who has ever struggled to adapt to a new system while the ghost of the old one haunts their instincts.
• 1:06
  SPEAKER_00: Welcome to Mindcast.
• 1:08
  SPEAKER_00: I'm Will, and this is the show where we take compelling ideas from cutting-edge research and translate them into something you can actually use.
• 1:17
  SPEAKER_00: Today's episode is built around research on mental model inertia.
• 1:21
  SPEAKER_00: What happens when the brain clings to an old internal map while operating in a fundamentally different world?
• 1:29
  SPEAKER_00: The paper is dense, but the insight is one of the most practically urgent I've encountered in years.
• 1:36
  SPEAKER_00: Here's the promise: by the time this episode ends, you'll understand why so much technology fails, not because of bad engineering, but because of the human mind's stubborn loyalty to what used to work.
• 1:49
  SPEAKER_00: You'll see it in hospitals, fighter jet cockpits, and corporate software rollouts, and you'll walk away with three concrete frameworks for navigating these transitions smarter.
• 2:00
  SPEAKER_00: Let's get into it.
• 2:03
  SPEAKER_00: Key insight number one the cognitive architecture of inertia.
• 2:09
  SPEAKER_00: What actually is a mental model?
• 2:12
  SPEAKER_00: In 1943, Scottish psychologist Kenneth Crake proposed something radical.
• 2:18
  SPEAKER_00: The human mind doesn't perceive reality directly.
• 2:21
  SPEAKER_00: Instead, it builds an internal simulation, a working model.
• 2:26
  SPEAKER_00: Think of it less like a camera and more like an architect's scale model of a city, approximate, useful, and deeply dependent on the assumptions baked in from the start.
• 2:37
  SPEAKER_00: In complex systems, hospital EHR platforms, aircraft cockpits, enterprise software, three distinct mental models are always in play.
• 2:48
  SPEAKER_00: The system model, how the technology actually works, the designer model, how engineers assumed users would interact with it, and the user mental model, the internal picture the actual human has built from experience and prior systems.
• 3:04
  SPEAKER_00: When all three align, technology feels intuitive.
• 3:08
  SPEAKER_00: When they diverge, and they almost always diverge during transition, you get friction, errors, and sometimes catastrophe.
• 3:16
  SPEAKER_00: The research asks, why is it so hard for the user's mental model to update when the system changes?
• 3:23
  SPEAKER_00: The answer is what researchers call mental model inertia.
• 3:29
  SPEAKER_00: Mental model inertia is the brain's tendency to preserve its existing internal simulation even when confronted with evidence that the world has changed.
• 3:40
  SPEAKER_00: It's not laziness, it's a cognitive survival mechanism.
• 3:44
  SPEAKER_00: Rebuilding a mental model from scratch takes energy and focus you often don't have in a high-stakes situation.
• 3:52
  SPEAKER_00: The research identifies four psychological mechanisms driving this inertia.
• 3:58
  SPEAKER_00: First, the Einstein effect.
• 4:01
  SPEAKER_00: Chess researchers in the 1940s found that expert players lock on to the first solution that comes to mind, even when a better one is right there.
• 4:11
  SPEAKER_00: Expertise becomes a cage.
• 4:14
  SPEAKER_00: Apply that to a veteran surgeon adapting to a new EHR system.
• 4:19
  SPEAKER_00: Years of experience paradoxically slow their adaptation.
• 4:23
  SPEAKER_00: Second, functional fixedness.
• 4:26
  SPEAKER_00: Seeing a system only in terms of its traditional function.
• 4:30
  SPEAKER_00: In systems transitions, this means treating digital databases as if they were the filing cabinets they replaced.
• 4:38
  SPEAKER_00: The form changed, the mental category did not.
• 4:42
  SPEAKER_00: Third, cognitive inertia, the resistance of mental schemas to updating.
• 4:48
  SPEAKER_00: Piaget's two modes, assimilation, filtering new information through existing schemas, and accommodation, where schemas are rebuilt.
• 4:58
  SPEAKER_00: Cognitive inertia is the preference for assimilation even when accommodation is needed.
• 5:05
  SPEAKER_00: Fourth, structural skewomorphism.
• 5:09
  SPEAKER_00: When a new digital system embeds the workflows of the legacy system it replaced, it becomes a digital photocopy of the old one, and neither users nor designers notice.
• 5:21
  SPEAKER_00: Tying all of this together, Hollenagel and Wood's joint cognitive systems framework.
• 5:27
  SPEAKER_00: The unit of analysis is the human technology pairing as a single integrated system.
• 5:33
  SPEAKER_00: When the technology half changes, the cognitive half doesn't automatically follow.
• 5:39
  SPEAKER_00: That gap is where things go wrong.
• 5:43
  SPEAKER_00: Key insight number two, the uncomfortable truth about legacy thinking.
• 5:48
  SPEAKER_00: At this point, you might think mental model inertia is the enemy.
• 5:53
  SPEAKER_00: We need people to abandon their old maps fast.
• 5:56
  SPEAKER_00: But the paper pushes back, and I think it's right to.
• 6:00
  SPEAKER_00: The researchers introduce carbon copy thinking, simultaneously a design liability and an evolutionary prerequisite.
• 6:09
  SPEAKER_00: Legacy mental models don't just carry baggage, they carry wisdom.
• 6:14
  SPEAKER_00: They encode thousands of hours of pattern recognition, failure modes avoided, intuitions earned.
• 6:20
  SPEAKER_00: When a hospital transitions to electronic records, the physician's mental model of what a clinical note should accomplish, communicate the patient's story, flag critical changes, support the next provider, that's crucial domain knowledge.
• 6:36
  SPEAKER_00: The problem arises when the form of the old system gets copied instead of the function.
• 6:42
  SPEAKER_00: The research identifies three ways legacy models are genuinely beneficial during transitions.
• 6:49
  SPEAKER_00: First, cognitive load offloading.
• 6:52
  SPEAKER_00: Familiar patterns act as anchors that free up bandwidth for genuinely new information.
• 6:58
  SPEAKER_00: Second, algorithmic trust calibration.
• 7:02
  SPEAKER_00: Experienced operators recognize when an algorithm is behaving suspiciously, they have a baseline.
• 7:08
  SPEAKER_00: Newer operators may defer to whatever the machine says.
• 7:12
  SPEAKER_00: Third, fail-safe generation.
• 7:15
  SPEAKER_00: Veterans understand the principles beneath the digital interface.
• 7:19
  SPEAKER_00: When it breaks down, they fall back to first principles problem solving in ways that operators trained only on the new system cannot.
• 7:28
  SPEAKER_00: The goal is to selectively migrate the wisdom while deliberately discarding the outdated form.
• 7:36
  SPEAKER_00: Key insight number three.
• 7:39
  SPEAKER_00: Three industries, three stories, one pattern.
• 7:43
  SPEAKER_00: The authors looked across commercial aviation, healthcare, and military defense and found the same pattern in each, with both devastating and unexpectedly positive results.
• 7:54
  SPEAKER_00: Start with aviation.
• 7:56
  SPEAKER_00: The transition from analog cockpits to what the industry calls glass cockpits, flight decks where the rows of individual mechanical dials pilots had relied on for decades were replaced by large integrated electronic display screens introduced a serious problem.
• 8:13
  SPEAKER_00: These aircraft could operate in dozens of different modes, each with subtly different behaviors, and those modes were not always clearly communicated to pilots.
• 8:23
  SPEAKER_00: Human factors researchers call this mode confusion.
• 8:27
  SPEAKER_00: Return to Flight 140.
• 8:30
  SPEAKER_00: A junior pilot accidentally engages go-around mode during approach.
• 8:34
  SPEAKER_00: The automation begins pushing the nose up.
• 8:37
  SPEAKER_00: The captain, whose mental model was built on thousands of hours in an older system, counters by pushing forward on the control column, not realizing this automated mode is extraordinarily resistant to manual override unless explicitly disengaged.
• 8:54
  SPEAKER_00: His instinct, right for the old system, was wrong in the new one.
• 8:58
  SPEAKER_00: The aircraft stalled.
• 9:03
  SPEAKER_00: But here's the nuance.
• 9:05
  SPEAKER_00: The same fly-by-wire automation that caused mode confusion also introduced flight envelope protection, a hard boundary around safe flight parameters the system maintains regardless of what the pilot commands.
• 9:19
  SPEAKER_00: In older aircraft, a disoriented pilot could push past structural limits.
• 9:24
  SPEAKER_00: In fly-by-wire aircraft, the automation refuses.
• 9:28
  SPEAKER_00: That feature is estimated to have prevented numerous other accidents.
• 9:32
  SPEAKER_00: The same mental model gap that killed in one scenario provided a safety net in another.
• 9:39
  SPEAKER_00: This is the dual role principle.
• 9:42
  SPEAKER_00: Now, healthcare.
• 10:52
  SPEAKER_00: Pick list errors have been linked to adverse drug events, real patient harm.
• 10:58
  SPEAKER_00: But EHRs also created something paper records never could: a comprehensive, time-stamped, immutable audit trail, invaluable in malpractice investigations and quality reviews.
• 11:09
  SPEAKER_00: The same rigidity that creates bloat provides legal clarity.
• 11:14
  SPEAKER_00: Same phenomenon, dual role.
• 11:16
  SPEAKER_00: Again.
• 11:38
  SPEAKER_00: The F-35 introduced sensor fusion.
• 11:41
  SPEAKER_00: Instead of raw data from individual sensors, the aircraft's computers synthesize everything into a single integrated tactical picture.
• 11:51
  SPEAKER_00: In theory, extraordinary.
• 11:53
  SPEAKER_00: In practice, many experienced pilots hated it and demanded access to the raw radar data underneath.
• 12:00
  SPEAKER_00: Why?
• 12:01
  SPEAKER_00: Their legacy mental models, built on years of interpreting raw sensor returns, depended on that raw information.
• 12:10
  SPEAKER_00: When the machine handed them a pre-digested answer, something felt wrong.
• 12:15
  SPEAKER_00: Not because the answer was wrong, but because their mental model couldn't validate it without seeing the underlying inputs.
• 12:23
  SPEAKER_00: The ODA loop, observe, orient, decide, act, is the foundational cognitive model for military decision-making.
• 12:32
  SPEAKER_00: When sensor fusion compresses observe and orient into a single pre-processed output, it disrupts the pilot's internalized ODA rhythm.
• 12:43
  SPEAKER_00: The orientation process, where experience and intuition turn raw data into situational awareness, gets short-circuited.
• 12:52
  SPEAKER_00: But the research documents instances where human operators caught what the algorithm missed.
• 12:58
  SPEAKER_00: Pilots noticed the fused picture had made a track swap, incorrectly associating radar returns between two different targets.
• 13:08
  SPEAKER_00: A human, drawing on years of pattern recognition, caught the error precisely because their mental model generated an anomaly flag.
• 13:17
  SPEAKER_00: The machine had no such flag.
• 13:19
  SPEAKER_00: It was confident and it was wrong.
• 13:22
  SPEAKER_00: The same legacy model that resists the new system is the layer that catches its errors, an irreplaceable feature of the human-machine partnership, if you design for it.
• 13:36
  SPEAKER_00: Three takeaways you can actually use.
• 13:39
  SPEAKER_00: Takeaway number one, reject the digitization of legacy flaws.
• 13:44
  SPEAKER_00: Use first principles thinking.
• 13:47
  SPEAKER_00: When leading a transition, the single most dangerous question is how do we replicate what we were doing before, but digitally?
• 13:55
  SPEAKER_00: That locks in every inefficiency and bottleneck of the legacy system, structural skew morphism by design.
• 14:03
  SPEAKER_00: The better question is, what is this process actually trying to accomplish?
• 14:08
  SPEAKER_00: Not what does it look like, what is the job it needs to do?
• 14:12
  SPEAKER_00: This is where Clayton Christensen's jobs to be done framework becomes enormously useful.
• 14:18
  SPEAKER_00: Applied to clinical documentation, instead of asking, how do we digitize the SOAP note, you ask, what is the job of a clinical note?
• 14:27
  SPEAKER_00: The answer, to accurately communicate a patient's current status, support safe decision making, and satisfy legal accountability requirements.
• 14:37
  SPEAKER_00: Once you frame it that way, copying 70% of the previous note is immediately a failure mode, not a feature.
• 14:45
  SPEAKER_00: A properly designed digital database can surface current vitals, flagged changes, and decision support alerts automatically.
• 14:53
  SPEAKER_00: The physician's cognitive energy is freed for actual clinical thinking, not transcription.
• 14:59
  SPEAKER_00: First principles thinking, combined with a clear jobs to be done framing, is the antidote to structural skewomorphism.
• 15:08
  SPEAKER_00: Takeaway number two.
• 15:10
  SPEAKER_00: Respect cognitive scaffolding.
• 15:12
  SPEAKER_00: Apply progressive skewomorphism.
• 15:14
  SPEAKER_00: This might sound like a contradiction of takeaway one, it isn't.
• 15:18
  SPEAKER_00: Rejecting legacy flaws is about what you design.
• 15:21
  SPEAKER_00: Respecting cognitive scaffolding is about how and when you deploy it.
• 15:26
  SPEAKER_00: The research proposes a three-phase model called progressive skewomorphism, converting something usually treated as a criticism into a deliberate, time-limited strategic tool.
• 15:37
  SPEAKER_00: Phase one, the scaffolding phase.
• 15:40
  SPEAKER_00: The new system presents itself in terms familiar to legacy users, not because the old design was better, but to lower the cognitive activation energy required to begin.
• 15:50
  SPEAKER_00: You meet users where they are.
• 15:52
  SPEAKER_00: The scaffolding is temporary and intentional, not permanent and accidental.
• 15:57
  SPEAKER_00: Phase 2, the hybrid phase.
• 15:59
  SPEAKER_00: New capabilities surface alongside familiar scaffolding.
• 16:03
  SPEAKER_00: Users who've stabilized on the basics begin encountering functions the legacy system could never provide.
• 16:09
  SPEAKER_00: They are invited, not forced, to explore.
• 16:12
  SPEAKER_00: Mental models stretch rather than snap.
• 16:15
  SPEAKER_00: Phase 3, the first principles phase, the scaffolding is removed.
• 16:19
  SPEAKER_00: The system operates on its native logic, designed from the ground up to exploit what the technology actually makes possible.
• 16:26
  SPEAKER_00: Users have cognitively migrated.
• 16:29
  SPEAKER_00: The key insight is sequencing.
• 16:31
  SPEAKER_00: Most organizations perform a big bang transition, flip the switch, and expect users at phase three from day one.
• 16:38
  SPEAKER_00: The cognitive science says that almost guarantees the very inertia and workarounds we've discussed.
• 16:44
  SPEAKER_00: Progressive skewomorphism acknowledges that cognitive migration takes time and structures that time deliberately.
• 16:52
  SPEAKER_00: Takeaway number three, design for the joint cognitive system.
• 16:56
  SPEAKER_00: Hollenagel and Wood's framework insists the unit of design is the human technology dyad as a single functioning cognitive system.
• 17:06
  SPEAKER_00: When you change one half, the other doesn't automatically update.
• 17:10
  SPEAKER_00: That gap is where accidents happen and where the primus of new technology goes unrealized.
• 17:16
  SPEAKER_00: Training programs and system designs need to account for the cognitive migration journey.
• 17:22
  SPEAKER_00: Not just here is how to use the new system, but here is how your existing knowledge maps onto it.
• 17:28
  SPEAKER_00: Here is where that mapping breaks down, and here is what you need to actively unlearn.
• 17:33
  SPEAKER_00: In aviation, this gave rise to crew resource management targeting mode confusion.
• 17:39
  SPEAKER_00: In healthcare, it points toward clinical informatics roles bridging physician mental models and EHR design.
• 17:46
  SPEAKER_00: In defense, it argues for tiered autonomy where human oversight is a deliberate feature, not a limitation.
• 17:53
  SPEAKER_00: The F-35 pilots who wanted raw radar data weren't being Luddites, they were expressing a legitimate cognitive need to maintain their ability to orient, validate, and remain genuinely in command, not passengers of the algorithm.
• 18:08
  SPEAKER_00: Designing for that need honors both the machine's computational power and the human's irreplaceable pattern recognition wisdom.
• 18:17
  SPEAKER_00: The organizations that manage transitions best stopped asking how do we get humans to adapt to the system and started asking how do we design the system to support the human's cognitive transition?
• 18:30
  SPEAKER_00: That reframe is everything.
• 18:34
  SPEAKER_00: Let me leave you with the thread running through all of this.
• 18:38
  SPEAKER_00: The human brain is not a computer you can reprogram with a software update.
• 18:43
  SPEAKER_00: It is a deeply adaptive, fundamentally conservative biological system that built its current operating model through years of lived experience.
• 18:54
  SPEAKER_00: That model deserves respect, not because it's always right, but because it contains wisdom that's hard to replicate from scratch.
• 19:02
  SPEAKER_00: Mental model inertia is not a flaw in the people navigating these transitions.
• 19:08
  SPEAKER_00: It's a predictable, well-understood phenomenon with known mechanisms and designable solutions.
• 19:15
  SPEAKER_00: The question is whether the people building new systems will take the cognitive science seriously enough to design for it.
• 19:23
  SPEAKER_00: Reject the digitization of legacy flaws.
• 19:26
  SPEAKER_00: Respect cognitive scaffolding designed for the joint cognitive system.
• 19:32
  SPEAKER_00: Three principles, one core insight.
• 19:35
  SPEAKER_00: Technology transitions are not just engineering challenges, they are cognitive challenges.
• 19:41
  SPEAKER_00: Organizations that build that into their design philosophy are the ones that actually realize the promise of new technology.
• 19:49
  SPEAKER_00: This has been Mindcast.
• 19:51
  SPEAKER_00: I'm Will.
• 19:52
  SPEAKER_00: Subscribe wherever you get your podcasts, and if you found this valuable, please leave a review.
• 19:57
  SPEAKER_00: It genuinely helps more people find the show.
• 20:00
  SPEAKER_00: Thank you for listening.
• 20:01
  SPEAKER_00: Stay curious, stay rigorous, and I'll see you on the next episode.