Abstract
In this paper we investigate how inference to the best explanation (IBE) works in engineering science, focussing on the context of malfunction explanation. While IBE has gotten a lot of attention in the philosophy of science literature, few, if any, philosophical work has focussed on IBE in engineering science practice. We first show that IBE in engineering science has a similar structure as IBE in other scientific domains in the sense that in both settings IBE hinges on the weighing of explanatory virtues. We then proceed to show that, due to the intimate connection between explanation and redesign in engineering science, there is a further engineering domain-specific virtue in terms of which engineering malfunction explanations are evaluated, viz. the virtue of redesign utility. This virtue entails that the explanatory information offered by a malfunction explanation should be instrumental in predicting counterfactual dependencies in redesigned systems. We illustrate and elaborate these points in terms of a number of engineering examples, focussing in particular on the 2009 crash of Air France Flight 447. Our extension of analyses of IBE and explanation to engineering science practice offers new insights by identifying a new explanatory virtue in malfunction explanation: redesign utility.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
We are not concerned here with the philosophical project of spelling out what makes function and malfunction ascriptions justifiable. Such normative accounts abound in philosophical theories of function, but are irrelevant for our purposes here, viz. elaborating how IBE works in the engineering practice of explaining malfunctions.
We use the terms ‘engineering science’, ‘engineering’, ‘engineering practice’, and ‘engineering science practice’ interchangeably in this paper. All research practices of engineers that we refer to by these terms we consider to be scientific practices. This paper is thus about engineering science, not about engineering: we take engineering science to refer to research that is aimed at developing new technical artifacts or at improving them; engineering in turn refers to the activity of constructing technical artifacts, based on knowledge gathered by engineering scientists.
Plumbridge (2009) studies the case of “Tin whiskers”. When changing led for a Tin alloy (for environmental reasons), it created nanofibers (whiskers). Sometimes two whiskers from nearby plaques connect, creating a short-circuit. This problem could not have been predicted, since the problems that arise at smaller levels are not analogous to problems at a macro-level.
For instance, Yun et al. (2018) studied how a closed innovation strategy and unidirectional chain of command led to batteries’ malfunctioning in Samsung Galaxy Notes 7 in 2016, due to which the smart phones ignited in a number of cases, whereas Loveridge et al. (2018) delved deep into the chemistry and physics involved in the short-circuiting batteries.
See, e.g. Serrat, Olivier (2017).
There are a number of different proposals in the literature for understanding mechanistic explanations. We endorse a generic notion of mechanistic explanation that most parties in the debate will agree on: mechanistic explanations articulate the mechanisms, organized collections of entities and activities, that produce, underlie, or constitute explanandum phenomena (Machamer et al. 2000; Bechtel and Abrahamson 2005; Craver 2007; Glennan 2005, 2017; Illari and Williamson 2012).
Psillos (2002) argues that in the process of IBE, evidence is already taken into account, i.e., in the inference to a best explanation, evidence and its quality has already been weighed and considered.
We use the notion of understanding advanced by Ylikoski and Kuorikoski (2010), where understanding is taken to be an inferential ability over counterfactual situations: the ability to answer contrastive what-if-things-had-been-different questions relating values of the explanans to values of the explanandum, where answering more (and more important) counterfactuals implies better understanding. Having a better understanding will allow for better manipulations and, in an engineering context, better redesigns.
This light bulb example is used as a running example to demonstrate how an automated method for failure analysis outputs malfunction explanations in the domain of (unmanned) autonomous aircraft in which these explanations are also connected to redesign issues (Snooke and Price 2012). The explanations generated by the method for instance are used to determine “whether all system failures can be observed with the existing sensors, and indicates which component failures can be isolated with existing sensors” (p. 871). In other words, explanations may reveal that a redesign of sensors is required. In line with this, part of the method is a model-generated FMEA by which “the consequences of every failure are reported to the designers, and they can decide the steps needed to improve the safety and reliability of the system.” (p. 871). If, say, an explanation for malfunctioning light bulbs indicates that the wiring connecting the light bulbs of a warning system to an aircrafts’ sub system for measuring altitude is prone to fracture or the switches in the circuitry are prone to corrosion, this may drive a redesign of parts of the circuitry – e.g., a different spatial configuration of the wiring or a novel type of switch. Redesign features as a component in the automated method for failure analysis from which this light bulb example derives.
During the summer of 2016 a number of Samsung Galaxy Note 7 phones were catching on fire. Samsung’s engineers promptly concluded that this had to do with faulty SDI batteries produced by a subsidiary supplier (Samsung SDI co.). However, the new supplier also started having issues with the batteries shortly after. Samsung hence declared: “We recognised that we did not correctly identify the issue the first time and remain committed to finding the root cause.” (2018: 271). By the 23th of January 2017 they gave a press conference where they explained what had happened together with several independent organizations.
Some other speculations were offered, but quickly disregarded because of their implausibility. For instance, a hail storm that would have caused a broken windshield (but then the pilots would have reported distress); a terrorist attack, which would explain the mysterious disappearance without a mayday call (but had there been an explosion, they would not have identified so many bodies in the debris). Highjack was also ruled out given the lack of radio warning and missile attack was disregarded as a possible explanation because of the altitude at which the plane flew. These explanatory hypotheses were discarded largely on evidential grounds.
An aerodynamic stall occurs when the wings do not produce enough lift, either because of insufficient speed or a wrong angle of attack. This is a very dangerous situation where the plane loses altitude quickly.
The explanations provided by the BBC and NOVA team are bona fide engineering explanations. The BBC and NOVA team were led by accredited experts and thus reflect engineering thinking and expertise on malfunction explanation. These experts included engineering failure analysts, aviation safety consultants, and aerospace and aviation meteorologists (see e.g., Williams et al. (2009) in which Air France Flight 447 is also investigated or Craig et al. (2008) on the detection of in-cloud turbulence). The documentary itself is referenced in the books “The Rio/Paris Crash: Uncovering the Secrets that Changed Aviation History” (Rapoport 2011) and “Understanding Air France 447” (Palmer 2013). The latter recovers a few threads of the documentary, making certain things more precise, such as the fact that supercooled water droplets likely adhered to snow crystals (forming graupels); and also giving further evidence to support this claim.
Traufetter (2010) gave a quite similar account of what transpired during the crash.
Pure water keeps its liquid form at temperatures well below 0C°, but if it touches impurities ice crystals rapidly form.
A contributory factor appears to have been the sophistication of the computerised flight controls. After autopilot disengaged, the pilots were bombarded with an array of information (more than 24 faults in under 4 min, sometimes inconsistent), which did not seem reliable with other measurements (e.g. speed would have read something like 60 knots). As the official report put it, the crew were over-saturated with what seemed like erroneous information. This is not the first incident of this kind -- Thomson (2013) showed that there are many cases where a poor interface hinders proper actions.
There were in fact 3 pilots. At the critical moment, the 2 co-pilots (Robert and Bonin) were piloting, while the captain (Dubois) was sleeping. The captain came into the cockpit once it was too late. We know this because he was not wearing a seatbelt (his body was found amongst the debris). The black boxes also confirm this.
Many have pointed out that in alternate law stall protection does not exist, but pilots are often not aware of this, and they probably believed the fly-by-wire system would not allow for such a problem (and that it was a malfunction of the signals).
In this paper we focus on explanations for malfunctions that, inter alia, cite faulty designs as causal factors for malfunctioning. Let us stress that the outcome of the explanations is that bad or suboptimal design is a source of the failures. Given such an outcome, we argue, an explanation must be formulated such that it has utility for redesign purposes. It is not the case that the explanatory projects from the outset presuppose faulty designs as causes of the malfunctions. Otherwise, our argument that these explanations should be evaluated in terms of their redesign utility would have a gloss of circularity. That is not the case. We thank a reviewer for urging us to clarify this matter.
Furthermore, redesign is noticeably driven by pragmatic considerations; which is a further difference from the other sciences, who’s models generally do not evolve according to pragmatic considerations (such as environmentally friendlier options).
In Woodward’s framework, more precisely, the explanatory generalization that figures in the explanans of the explanation.
Our analysis is complementary to the one of Weber et al. (2019). We identify virtues that malfunction explanations should meet, whereas Weber et al. (2019) assess the utility of predictions in redesign contexts. The source of these predictions can be failure analyses. So, the class of malfunction explanations we consider – those that should meet the virtue of redesign utility – should offer information that is instrumental in predicting dependencies in to-be redesigned systems. These explanations thus, as we argue, should offer information to formulate and answer what-would-happen-if-questions. Weber et al. (2019) assess the structure and utility of these predictive answers.
References
Affonso, L.O.A. (2007). Machinery failure analysis handbook. Gulf Publishing Company.
Amaya, A. (2007). Inference to the best legal explanation. SSRN Electronic Journal.
BEA (2012). Final report on the accident on 1st June 2009 to the Airbus A330–203 registered F-GZCP operated by air France flight AF 447 Rio de Janeiro – Paris (pdf), translated by BEA from French, Le Bourget: BEA Bureau d'Enquêtes et d'Analyses pour la sécurité de l'aviation civile, retrieved 15 May 2019.
Bechtel, W., & Abrahamson, A. (2005). Explanation: A mechanist alternative. Studies in History and Philosophy of Biological and Biomedical Sciences, 36, 421–441.
Bhaumik, S. (2009). A view on the general practice in engineering failure analysis. Journal of Failure Analysis and Prevention, 9(3), 185–192.
Bird, A. (2010). Eliminative abduction: Examples from medicine. Studies in History and Philosophy of Science, 41, 345–352.
Birolini, A. (2007). Reliability engineering: Theory and practice (5th ed.). Berlin: Springer.
Boukharouba, T., Elboujdaini, M., & Pluvinage, G. (Eds.) (2009). Damage and fracture mechanics. Failure analysis of engineering materials and structures. Springer.
Craig, J. A., Williams, J.K., Blackburn, G., Linden, S. and Stone R. (2008). Remote detection and real-time alerting for in-cloud turbulence. AMS 13th Conference on Aviation, Range and Aerospace Meteorology, 10.5.
Craver, C. F. (2007). Explaining the brain: Mechanisms and the mosaic unity of neuroscience. New York: Oxford University Press.
Darlow Smith productions, & Scott, K. (2010). Lost: The mystery of flight 447. London: BBC.
de Frate, L. (2014). Failure: Analysis of an engineering concept. Philosophy of Technology.
Dragulinescu, S. (2016). Inference to the best explanation and mechanisms in medicine. Theoretical Medicine and Bioethics, 37(3), 211–232.
Dragulinescu, S. (2017). Inference to the best explanation as a theory for the quality of mechanistic evidence in medicine. European Journal for Philosophy of Science, 7(2), 353–372.
Feld, J., & Carper, K. (1997). Construction failure (2nd ed.). New York: Wiley.
Fricker, E. (2017). Inference to the best explanation and the receipt of testimony: Testimonial reductionism vindicated. In (Ed.), Best Explanations: New Essays on Inference to the Best Explanation. Oxford University press.
Glennan, S. (2005). Modeling mechanisms. Studies in History and Philosophy of Biological and Biomedical Sciences, 36(2), 375–388.
Glennan, S. (2017). The new mechanical philosophy. Oxford University Press.
Hoffmann, M. H. (2010). “Theoric transformations”and a new classification of abductive inferences. Transactions of the Charles S. Peirce Society: A Quarterly Journal in American Philosophy, 46(4), 570–590.
Houkes, W., & Vermaas, P. E. (2010). Technical functions: On the use and design of artefacts (Vol. 1). Springer Science & Business Media.
Illari, P., & Williamson, J. (2012). What is a mechanism? Thinking about mechanisms across the sciences. European Journal for Philosophy of Science, 2, 119–135.
Jespersen, B., & Carrara, M. (2013). A new logic of technical malfunction. Studia Logica, 101(3), 547–581.
Josephson, J. R., & Josephson, G. S. (Eds.). (1994). Abductive inference. Cambridge: Cambridge university Press.
Krohs, U. (2009). Functions as based on a concept of general design. Synthese, 166, 69–89.
Lipton, P. (2004). Inference to the best explanation. Taylor and Francis Group: Routledge.
Loveridge, M. J., Remy, G., Kourra, N., Genieser, R., Barai, A., Lain, M. J., et al. (2018). Looking deeper into the Galaxy (Note 7). Batteries, 4(1), 3.
Machamer, P. K., Darden, L., & Craver, C. F. (2000). Thinking about mechanisms. Philosophy of Science, 57, 1–25.
Niiniluoto, I. (1999). Defending abduction. Philosophy of Science, 66, S436–S451.
Palmer, W. (2013). Understanding air France 447. Paperback.
Plumbridge, W. J. (2009). New avenues for failure analysis. Engineering Failure Analysis, 16(5), 1347–1354.
Preston, B. (1998). Why is a wing like a spoon? A pluralist theory of functions. Journal of Philosophy, 95, 215–254.
Psillos, S. (2002). Causation and explanation. Acumen & McGill-Queens U.P.
Rapoport, R. (2011). The Rio/Paris crash: Uncovering the secrets that changed aviation history. James Sparling.
Samsung (2017). Samsung Electronics Announces Cause of Galaxy Note7 Incidents in Press Conference [Press release].
Schupbach, J. N. (2017). Inference to the best explanation, cleaned up and made respectable". Best Explanations: New Essays on Inference to the Best Explanation. pp. 39–61.
Schurz, G. (2008). Patterns of abduction. Synthese, 164(2), 201–234.
Serrat, O. (2017). The five whys technique. In Knowledge solutions (pp. 307–310). Singapore: Springer.
Snooke, N., & Price, C. J. (2012). Automated FMEA based diagnostic symptom generation. Advanced Engineering Informatics, 26, 870–888.
Thagard, P. R. (1978). The best explanation: Criteria for theory choice. Journal of Philosophy, 75(2), 76–92.
Thomson, J. (2013). Situation awareness and the human-machine interface. Safety in Engineering Ltd.
Traufetter, G. (2010). Death in the Atlantic: The last four minutes of air France flight 447. Spiegel Online. Retrieved 15 May 2019.
UL. (2017). FAILURE ANALYSIS OF SAMSUNG NOTE 7 [Press release].
van Eck, D. (2015). Mechanistic explanation in engineering science. European Journal for Philosophy of Science, 5(3), 349–375.
van Eck, D. (2016). The philosophy of science and engineering design. Springer.
van Eck, D., & Weber, E. (2014). Function ascription and explanation: Elaborating an explanatory utility desideratum for ascriptions of technical functions. Erkenntnis., 79(6), 1367–1389.
Van Eck, D., & Weber, E. (2017). In defense of coexisting engineering meanings of function. AI EDAM, 31(1), 55–68.
Weber, E., van Eck, D., & Mennes, J. (2019). On the structure and epistemic value of function ascriptions in biology and engineering sciences. Foundations of Science, 24(3), 559–581.
Weslake, B. (2010). Explanatory depth. Philosophy of Science, 77(2), 273–294.
Williams, J. K., Sharman, R. and Kessinger, C. (2009). Developing a global turbulence and convection Nowcast and forecast system. Published by the American Institute of Aeronautics and Astronautics.
Woodward, J. (2003). Making things happen: A theory of causal explanation. Oxford University Press.
Woodward, J., & Hitchcock, C. (2003). Explanatory generalizations, part I: A counterfactual account. Noûs, 37(1), 1–24.
Ylikoski, P., & Kuorikoski, J. (2010). Dissecting explanatory power. Philosophical Studies, 148(2), 201–219.
Yun, J. J., et al. (2018). Benefits and Costs of Closed Innovation Strategy: Analysis of Samsung’s Galaxy Note 7 Explosion and Withdrawal Scandal. Journal of Open Innovation: Technology, Market, and Complexity, 4(3), 20.
Acknowledgements
We thank Rami Koskinen, Erik Weber, and two anonymous referees for useful feedback on earlier versions of this paper.
Funding
Funding was provided to Kristian González Barman and Dingmar van Eck by the Research Foundation Flanders (FWO).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare that they have no conflicts of interest.
Ethical approval
Not applicable.
Informed consent
Not applicable.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Barman, K.G., van Eck, D. IBE in engineering science - the case of malfunction explanation. Euro Jnl Phil Sci 11, 10 (2021). https://doi.org/10.1007/s13194-020-00325-6
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s13194-020-00325-6