← Lucian Cosinschi

DIGEST — A8: State of the Art, 2015–August 2026

Digest (~900 words)

The reform became law, and the law created a measurement problem it has not solved. Competency-based training stopped being a movement in this period and became a compliance regime. Mixed EBT has been permitted in Europe since ED Decision 2015/027/R; baseline EBT entered the hard rules as ORO.FC.231 via Commission Implementing Regulation (EU) 2020/2036, with AMC/GM in ED Decision 2021/002/R. EASA’s oversight guidance (RMT.0599 safety material, v1.3, 2Q 2025) sets an approval ladder that requires three years of mixed EBT plus two years of a demonstrated instructor concordance assurance programme before baseline EBT — a six-year gate that alone explains why baseline EBT is still uncommon in 2026. Crucially, EASA names the psychometric target explicitly (“maximising instructor’s concordance (inter-rater reliability)”) and requires authorities to verify “the accuracy of the grading system” — but sets no numeric threshold. An operator satisfies the rule by having a concordance programme, not by reaching concordance. Meanwhile no operator has published inter-rater reliability figures: Lufthansa’s ICAP (EATS 2024) is described in method and rationale with no kappa, ICC or agreement percentage. The only high-agreement result available (Sun et al. 2023, r = 0.947 vs expert panel) was obtained by removing the human rater and scoring flight-path parameters algorithmically — i.e. by narrowing the construct back to the tolerance-based checking CBTA was designed to replace. That is the central irony of the current state of the art.

Startle and surprise: a well-evidenced problem, a thin solution, and now partial disconfirmation. Landman et al.’s 2017 conceptual model (frames, sensemaking, reframing) licensed the whole training category. Landman et al. (2017b) showed the same stall flown worse when unexpected (all four FAA recovery criteria, d = 0.61–1.23) — establishing the problem. Landman et al. (2018) is the only study testing an intervention: 9/10 vs 2/10 successful landings for unpredictable/variable practice (χ² = 9.90, p = .002), n = 10 per cell — and, critically, no group difference on the unrelated surprise test and none on manual skill. There is no evidence of general, transferable “startle resistance.” Then Chen et al. (2026, Human Factors) quantified startle at ~49 ms of added reaction time per scale point but found surprise had no significant association with information-processing performance — a negative result against the field’s own dominant model, from the model’s own group. The mechanism that works is scenario variability, which is an argument for unpredictable training design generally, not for a bolt-on startle module.

VR/XR crossed the regulatory line decisively; the evidence did not follow. The qualification record is short and specific: EASA qualified the first VR FSTD on 26 April 2021 — an FNPT II for the Robinson R22 Beta II, built by VRM Switzerland (now Loft Dynamics), under Special Conditions because no CS covered VR. FAA qualification of the H125 device followed 31 July 2024 (14 CFR Part 60, at Marshall University), then FAA Level 7 FTD on 30 September 2025. France’s DGAC approved the first OEM VR-based type-rating programme (Airbus Helicopters, AS350 B3e) on 8 May 2025. The first EASA-qualified mixed-reality device — and the first fixed-wing one — was BRUNNER’s NOVASIM MR DA42 (FNPT II, Varjo XR-4, at Lufthansa Aviation Training), announced 4 June 2025, with integrated eye tracking. No VR/XR device is qualified for a large transport aeroplane type. The structural change came on 15 July 2026: Regulation (EU) 2026/781 and CS-FSTD Issue 1 replace fixed device types and levels with a capability-based system built on the FSTD Capability Signature, with explicit qualification provisions for XR and touchscreen flight decks and an optional task-to-tool methodology (applicability reportedly 30 April 2028). Against that: the published aviation VR effectiveness literature is a handful of small quasi-experiments, and the best of them (Frontiers, 2026) found VR-trained ab-initio students flew their first real flight better but learned theory worse than a classroom control.

AI is deployed in the debrief layer, not the assessment layer. EASA’s framework (AI Roadmap 1.0, Feb 2020; 2.0, May 2023; Concept Paper Issue 2, 6 Mar 2024; Proposed Issue 3, 3 Jun 2026) is elaborate for AI in the aircraft — and essentially silent on AI assessing pilots. Roadmap 2.0’s only substantive training statement is one sentence about adaptive training. Navi AI (public 25 Mar 2026, deploying to Embry-Riddle) and the AXIS AI Debriefing Station (22 Dec 2025) summarise, tag and flag; neither is the authority on a licence-relevant check anywhere I could find.

eMCO stalled — on EASA’s own evidence. The NLR-led consortium’s D-9 final report (June 2025, EASA.2022.C17) concluded that “an equivalent level of safety between eMCO and normal crew operations can currently not be demonstrated,” blocked by incapacitation detection, sleep inertia (~35 min), loss of cross-checking, and physiological constraints. ECA’s deepest training objection is the apprenticeship one: command competence is transmitted by co-presence with captains.

And a structural contradiction worth building on. EASA’s SIB 2013-05R1 (23 June 2025) recommends operators use FDM to monitor the consequences of manual flying — while the IDT/Advani–Abbott survey presented at EATS 2025 reports that fear of FDM flagging is precisely what stops pilots practising, with the effect strongest among the least experienced. EASA also states outright that it disagrees with FAA AC 120-123’s more permissive stance on extensive manual flight in line operations. The monitoring apparatus built to detect degraded handling is a cause of it.

18 strongest citations

  1. Landman, A., Groen, E. L., van Paassen, M. M., Bronkhorst, A. W., & Mulder, M. (2017). “Dealing With Unexpected Events on the Flight Deck: A Conceptual Model of Startle and Surprise.” Human Factors, 59(8), 1161–1172. https://doi.org/10.1177/0018720817723428 [VERIFIED]

  2. Landman, A., Groen, E. L., van Paassen, M. M., Bronkhorst, A. W., & Mulder, M. (2017). “The Influence of Surprise on Upset Recovery Performance in Airline Pilots.” The International Journal of Aerospace Psychology, 27(1–2), 2–14. https://doi.org/10.1080/10508414.2017.1365610 [VERIFIED — full text via TNO repository]

  3. Landman, A., van Oorschot, P., van Paassen, M. M., Groen, E. L., Bronkhorst, A. W., & Mulder, M. (2018). “Training Pilots for Unexpected Events: A Simulator Study on the Advantage of Unpredictable and Variable Scenarios.” Human Factors, 60(6), 793–805. https://doi.org/10.1177/0018720818779928 [VERIFIED]

  4. Chen, J., Landman, A., Derumigny, A., Stroosma, O., van Paassen, M. M., & Mulder, M. (2026). “Relationships Between Pilots’ Startle and Surprise Responses and Information-Processing Performance During Simulated In-Flight Events.” Human Factors, 68(8). https://doi.org/10.1177/00187208261434426 [VERIFIED]

  5. Rivera, J., Talone, A. B., Boesser, C. T., Jentsch, F., & Yeh, M. (2014). “Startle and Surprise on the Flight Deck: Similarities, Differences, and Prescriptions.” Proceedings of the Human Factors and Ergonomics Society Annual Meeting, 58(1), 1047–1051. https://doi.org/10.1177/1541931214581219 [VERIFIED — citation/venue; full text not read]

  6. EASA (2025). Safety Material RMT.0599 ‘Evidence-based and competency-based training’ — Oversight guidance for the transition to EBT Implementation Baseline, v1.3, 2Q 2025. https://www.easa.europa.eu/en/downloads/131390/en [VERIFIED — read directly]

  7. IATA. Competency Assessment and Evaluation for Pilots, Instructors and Evaluators — Guidance Material. https://www.iata.org/contentassets/c0f61fc821dc4f62bb6441d7abedb076/competency-assessment-and-evaluation-for-pilots-instructors-and-evaluators-gm.pdf [VERIFIED]

  8. Sun, H., Zhou, X., Zhang, P., Liu, X., Lu, Y., Huang, H., & Song, W. (2023). “Competency-based assessment of pilots’ manual flight performance during instrument flight training.” Cognition, Technology & Work, 25, 345–356. https://doi.org/10.1007/s10111-023-00737-3 [VERIFIED — full text read]

  9. EASA (2021). “EASA approves the first Virtual Reality (VR) based Flight Simulation Training Device.” Press release, Cologne, 26 April 2021. https://www.easa.europa.eu/en/newsroom-and-events/press-releases/easa-approves-first-virtual-reality-vr-based-flight-simulation [VERIFIED — FNPT II, Robinson R22 Beta II, Special Conditions]

  10. EASA (2026). “EASA modernises flight simulation training devices’ rules to support innovation and future pilot training.” Press release, Cologne, 15 July 2026 — Regulation (EU) 2026/781 and CS-FSTD Issue 1. https://www.easa.europa.eu/en/newsroom-and-events/press-releases/easa-modernises-flight-simulation-training-devices-rules-support-innovation-and-future-pilot [VERIFIED]

  11. Varjo / BRUNNER Elektronik AG (2025). “Varjo XR Technology Powers First-Ever EASA-Qualified Mixed Reality Flight Simulator by BRUNNER,” 4 June 2025 — NOVASIM MR DA42, FNPT II, Lufthansa Aviation Training. https://varjo.com/news/varjo-xr-technology-powers-first-ever-easa-qualified-mixed-reality-flight-simulator-by-brunner [SECONDARY]

  12. NLR for EASA (2025). eMCO-SiPO EASA.2022.C17 D-9 Final Report on Risk Assessment for eMCOs and SiPOs, June 2025. https://www.easa.europa.eu/en/downloads/142086/en — and companion D-8 Report on the Preliminary Safety Risk Identification for SiPOs, 5 June 2025, https://www.easa.europa.eu/en/downloads/142085/en [VERIFIED]

  13. European Cockpit Association (2025). The Human and the concepts of Reduced Crew Operations (RCO): Extended Minimum Crew Operations & Single Pilot Operations, Position Paper, January 2025. https://www.eurocockpit.eu/sites/default/files/2025-01/eMCO_SiPO_ECA_Position_PP_UPDATE_25_0109_F.pdf [VERIFIED]

  14. FAA (2020). Flight Standardization Board Report, Boeing 737, Revision 17, 16 November 2020 — Appendix 7, “Boeing 737 MAX Special Training for Flightcrews”; MDR annotation “(E*) B/B”. https://www.faa.gov/sites/faa.gov/files/2022-08/737_FSB_Report.pdf [VERIFIED]

  15. EASA (2025). Safety Information Bulletin 2013-05R1, “Manual Flight Training and Operations,” issued 23 June 2025. https://ad.easa.europa.eu/blob/EASA_SIB_2013_05_R1.pdf/SIB_2013-05R1_1 [VERIFIED]

  16. Causse, M., Mercier, M., Lefrançois, O., & Matton, N. (2025). “Impact of automation level on airline pilots’ flying performance and visual scanning strategies: A full flight simulator study.” Applied Ergonomics, 125, 104456. https://doi.org/10.1016/j.apergo.2024.104456 [VERIFIED — abstract via TRID; full text blocked]

  17. U.S. Government Accountability Office (2026). Aviation Workforce: FAA Could Strengthen Regional Pilot Pipeline by Establishing Timelines for Training Initiatives, GAO-26-107856, April 2026. https://files.gao.gov/reports/GAO-26-107856/index.html [VERIFIED]

  18. EASA / EOFDM (2024). Breaking the Silos, Revision 1, December 2024 — individual FDM summary reports; the “circle of confidence” doctrine. https://www.easa.europa.eu/sites/default/files/dfu/BreakingTheSilos_Revision1_Dec2024_0.pdf [VERIFIED]

Supporting, also verified: IATA, CBTA Expansion within the Aviation System, White Paper, March 2024; EASA AI Concept Paper Issue 2 (6 Mar 2024) and Proposed Issue 3 (3 Jun 2026); Halldale, “Global Survey Reveals Manual Flight Skills Crisis,” 23 Dec 2025 (IDT/Advani–Abbott survey at EATS 2025).

Deployed vs. announced vs. marketed

GENUINELY DEPLOYED (regulator-confirmed or institutionally in service)

  • Mixed EBT and baseline EBT as approved regulatory pathways in Europe; RMT.0599 oversight machinery in force.
  • Instructor concordance assurance programmes as a condition of baseline EBT approval (Lufthansa ICAP is a real programme).
  • Six VR/XR qualification events: EASA FNPT II R22 (26 Apr 2021); EASA H125 (late 2023, vendor-stated); FAA H125 (31 Jul 2024); DGAC AS350 B3e type rating (8 May 2025); EASA MR DA42 FNPT II (4 Jun 2025); FAA Level 7 H125 (30 Sep 2025).
  • Regulation (EU) 2026/781 + CS-FSTD Issue 1 — capability-based FSTD framework, published 15 July 2026.
  • Post-MAX mandatory FFS special training under FSB Rev 17 Appendix 7.
  • Individual FDM feedback dashboards to pilots (21 KPIs, monthly) under EOFDM-documented confidentiality architecture.
  • EASA SIB 2013-05R1 manual-flight guidance (non-mandatory).

ANNOUNCED OR PILOTED, NOT DELIVERED

  • Boeing 737 VR simulator (Alaska Airlines / Loft Dynamics, April 2025) — does not exist in qualified form; Alaska’s own wording is conditional. No airline trains 737 crews in VR today.
  • FAA Enhanced Qualification Program — legislated in the FAA Reauthorization Act of 2024 with a November 2024 deadline; as of February 2026 not issued and no timeline set.
  • eMCO/SiPO — research concluded equivalence cannot be demonstrated; ECA’s tentative timeline shows consultation 2027, Opinion 2029, decision 2030.
  • AI in competency assessment — no regulatory framework, no deployed licence-relevant use found.
  • AirTanker’s VR alternative to live aircraft visits — UK CAA approval sought, not granted.

VENDOR CLAIMS (not evidence)

  • CAE Rise standardisation and effectiveness claims; AXIS AI Debriefing Station claims to “standardise assessments across instructors”; Navi AI’s “100,000 real flight hours” corpus and Embry-Riddle deployment; Loft Dynamics’ “all aspects of pilot training and checking” scope for the 2023 EASA H125 qualification; Varjo’s citation of USAF data (“33 of 40 manoeuvres statistically significant”); American Airlines’ “substantial trainee performance improvements.”

Explicit gaps

  1. No published count of EBT-approved operators anywhere — EASA, national authorities or IATA. “Widespread EBT adoption” is unsupported by any register I could locate.
  2. No published evaluation showing EBT/CBTA produces a measurable safety outcome. After ~13 years of ICAO endorsement, the evidence-based programme has published no evidence about itself. This is the single biggest hole.
  3. No published inter-rater reliability figures from any operator concordance programme. No numeric threshold in EASA’s rule either.
  4. No published evaluation of ORCA (EASA’s recommended observe–record–classify–assess grading discipline) on rater reliability in aviation.
  5. Martin, Murray, Bates & Lee startle work not retrieved — full text inaccessible in this session. Do not cite from memory.
  6. ICAO A42 (Sep–Oct 2025) WP/521 on extended minimum crew operations not retrieved (ICAO server blocked automated access) and the Assembly’s disposition of it is unknown. URL: https://www.icao.int/sites/default/files/Meetings/a42/Documents/WP/wp_521_en.pdf
  7. IFALPA and ALPA position papers on reduced-crew operations not read (bot protection). Titles/hosts confirmed only.
  8. FAA’s 2012/2013 regulatory evaluation of the 1,500-hour rule not read directly — the “no quantifiable relationship” finding and the 35%/55% figures come via The Air Current and must be verified against the original (2013 final rule, 78 FR 42324).
  9. Pilot Source Study specific 2018 figure unverified — the body of work is real; the number needs sourcing to the original paper.
  10. No recent independent evaluation of MPL graduate performance versus traditional-route graduates located. This matters: MPL is the purest test of “competency not hours.”
  11. Applicability date of Regulation (EU) 2026/781 (reportedly 30 April 2028) comes from trade press, not the EASA press release. Verify against the Official Journal text.
  12. EASA’s 737 MAX return-to-service training requirements not compared with the FAA’s; no check on whether the “(E*)” device has been reused for other types.
  13. No guidance found anywhere addressing AI debrief tools’ interaction with FDM confidentiality — Navi ingests cockpit audio in training environments.
  14. Full text not obtained for: PETRA ‘25 automated competency assessment paper (10.1145/3733155.3733195); AIAA SciTech 2025 AI-generated feedback paper (10.2514/6.2025-2729); Ergonomics 68(12) 2025 simulator-fidelity study (10.1080/00140139.2024.2449110); Faulhaber et al. 2022 single-pilot scanning study. All four are worth retrieving.