See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/352061422

Explanation in AI systems

Chapter · June 2021

CITATIONS READS

0 46

2 authors:

Marko Tesic Ulrike Hahn

Birkbeck, University of London Birkbeck, University of London
7 PUBLICATIONS   15 CITATIONS    189 PUBLICATIONS   3,926 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Acquired Equivalence, Similarity, & Cross-Classification View project

Communication Across Social Networks View project

All content following this page was uploaded by Marko Tesic on 02 June 2021.

The user has requested enhancement of the downloaded file.

 Explanation in AI systems

Marko Tešić and Ulrike Hahn

Department of Psychological Sciences, Birkbeck, University of London

1
Acknowledgements
This work was supported by the Intelligence Advanced Research Projects Activity
(IARPA) as part of the Bayesian Argumentation via Delphi (BARD) project, and by
the Humboldt Foundation.
1
Explanation in AI systems

In this chapter, we consider recent work aimed at guiding the design of algorithmically
generated explanations. The chapter proceeds in four parts. Firstly, we introduce the
general problem of machine-generated explanation and illustrate different notions of
explanation with the help of Bayesian belief networks. Secondly, we introduce key
theoretical perspectives on what constitutes an explanation, and more specifically
a ‘good’ explanation, from the philosophy literature. We compare these theoretical
perspectives and the criteria they propose with a case study on explaining reasoning
in Bayesian belief networks and present implications for AI. Thirdly, we consider the
pragmatic nature of explanation with the focus on its communicative aspects that are
manifested in considerations of trust. Finally, we present conclusions.

1.1 Machine-generated explanation

Recent years have seen a groundswell of interest in machine-generated explanation for
AI systems (DARPA, 2016; Doshi-Velez and Kim, 2017; Montavon et al., 2018; Rieger
et al., 2018; Samek et al., 2017). Multiple factors exert pressure for supplementing
AI systems with explanations of their outputs. Explanations provide transparency
for what are often black-box procedures. Hence transparency is viewed as critical
for fostering the acceptance of AI systems in real-world practice (Bansal et al., 2014;
Chen et al., 2014; Fallon and Blaha, 2018; Hayes and Shah, 2017; Mercado et al., 2016;
Wachter et al., 2017b), last but not least, because transparency might be a necessary
ingredient for dealing with legal liability (Felzmann et al., 2019; Doshi-Velez et al.,
2017; Goodman and Flaxman, 2016; Wachter et al., 2017a). At the same time, decades
of research in AI make plausible the claim that AI systems genuinely able to navigate
real-world challenges are likely to involve joint human-system decision making, at least
for the foreseeable future. This however, requires AI systems to communicate outputs
in such a way as to allow humans to make informed decisions.
The challenge of developing adequate, machine-generated explanation is a formidable
one. For one, it requires an accessible model of how AI system’s outputs or conclusions
were arrived at. This poses non-trivial challenges for many of the presently most suc-
cessful types of AI systems, such as convolutional, deep learning networks (Collobert
et al., 2011; Goodfellow et al., 2016; Graves and Schmidhuber, 2005; Krizhevsky et al.,
2012) which are already notoriously opaque to external observers, let alone offering up
representations of a system’s internal processes that could be used to drive explana-
tion generation. As such, the automated generation of explanation is arguably easier
to achieve with AI systems that operate with models that are formulated at acceptable
2 Explanation in AI systems

user levels of engagement; at least here, the step of translating low-level representa-
tions into a suitable higher-level representations accessible to us is, in a large number
of cases, already taken care of.
Bayesian Belief Networks (BNs) are an AI technique that has been viewed as sig-
nificantly more interpretable and transparent than deep neural networks (Gunning
and Aha, 2019), while still possessing a notable predictive power and being applied
to various contexts ranging from defence and military (Falzon, 2006; Laskey and Ma-
honey, 1997; Lippmann et al., 2006) and cyber security (Chockalingam et al., 2017; Xie
et al., 2010), over medicine (Agrahari et al., 2018; Wiegerinck et al., 2013), and law
and forensics (Lagnado et al., 2013; Fenton et al., 2013), to agriculture (Drury et al.,
2017) as well as psychology, philosophy, and economics (see below). As such, BNs seem
to serve well one of the goals of this chapter which is to bring and overlay insights
on explanations from different areas of research: they are a promising meeting point
connecting the research on machine-generated explanation in AI and the research on
human understanding of explanation in psychology and philosophy. We thus use BNs
as the focal point of our analysis in this chapter. Given the increasing popularity of
BNs within AI (Friedman et al., 1997; Pernkopf and Bilmes, 2005; Roos et al., 2005;
Ng and Jordan, 2002), including their relation to deep neural networks (Choi et al.,
2019; Rohekar et al., 2018; Wang and Yeung, 2016) and efforts to explain deep neural
networks via BNs (Harradon et al., 2018), this should be intrinsically interesting. Fur-
thermore, we take the kinds of issues we identify here to be indicative of the kinds of
problems and distinctions that will likely emerge in any attempt at machine-generated
explanation.

1.1.1 Bayesian belief networks: a brief introduction

Bayesian belief networks provide a simple graphical formalism for summarising and
simplifying joint probability distributions in such a way as to facilitate Bayesian com-
putations (Neapolitan, 2003; Pearl, 1988). Specifically, BNs use independence relations
between variables to simplify the computation of joint probability distributions in cases
of multivariate problems. As Bayesian models, they have a clear normative foundation:
“being Bayesian”, that is, assigning degrees of belief in line with the axioms of proba-
bility (the Dutch Book argument, Ramsey 2016; Vineberg 2016), and the use of Bayes’
rule to update believes in light of new evidence (also known as ‘conditionalization’)
which uniquely minimises the inaccuracy of an agent’s beliefs across all possible worlds
(that is, regardless of how the world turns out), on the condition that inaccuracy is
measured with the Brier score and those worlds are finite (see, e.g., the formal re-
sults outlined in Pettigrew 2016). In other words, Bayesian computations specify how
agents should change their beliefs, if they wish those beliefs to be accurate. As a re-
sult, the Bayesian framework has seen widespread application not just in AI, but also
in philosophy, economics, and psychology (Bovens and Hartmann, 2003; Dardashti
et al., 2019; Dewitt et al., 2018; Dizadji-Bahmani et al., 2011; Fenton et al., 2013;
Hahn and Oaksford, 2006; Hahn and Oaksford, 2007; Hahn and Hornikx, 2016; Harris
et al., 2016; Howson and Urbach, 2006; Liefgreen et al., 2018; Madsen et al., 2018;
Neil et al., 2008; Phillips et al., 2018; Pilditch et al., 2018; Rehder, 2014; Rottman and
Hastie, 2014; Spiegler, 2016; Tešić, 2019; Tešić and Hahn, 2019; Tešić et al., 2020).
 Machine-generated explanation 3

In particular, BNs are helpful in spelling out the implications of less intuitive inter-
actions between variables. This is readily illustrated with the example of “explaining
away”, a phenomenon that has received widespread psychological investigation (Davis
and Rehder, 2017; Fernbach and Rehder, 2013; Liefgreen et al., 2018; Morris and Lar-
rick, 1995; Pilditch et al., 2019; Rehder, 2014; Rehder and Waldmann, 2017; Rottman
and Hastie, 2014; Rottman and Hastie, 2016; Sussman and Oppenheimer, 2011; Tešić
et al., 2020). Figure 1.1 illustrates a simple example of explaining away. There are two
potential causes, a physical abuse and haemophilia (a genetic bleeding disorder), of
a single effect, bruises on a child’s body. Before finding out anything about whether
there are bruises on a body, the two causes are independent: learning that a child
is suffering from haemophilia will not change our beliefs about whether the child is
physically abused. However, if we learn that the child has bruises on its body, then the
two causes become dependent: additionally learning that the child is suffering from
haemophilia will change (decrease) the probability that it has been physically abused
since haemophilia alone is sufficient to explain away the bruises.

Fig. 1.1 A BN model of explaining away.

The example illustrates not just BNs ability to model explaining away situations
and provide us with both qualitative and quantitative normative answers, but also
their advantage over classical logic and rule-based expert systems. A rule-based expert
system consisting of a set of IF-THEN rules and a set of facts (see Grosan and Abraham
2011) may carry out an incorrect chaining in situations representing explaining away.
For instance, a rule-based system may combine plausibly-looking rules “If the child is
suffering from haemophilia, then it is likely the child has bruises” with “If the child
has bruises, then it is likely the child is physically abused” to get “If the child is
suffering from haemophilia, then it is likely the child is physically abused”. However,
we know that actually the opposite is true: learning about haemophilia makes physical
abuse less likely (Pearl, 1988). The application of rule-based expert systems to legal
and medical contexts (Grosan and Abraham, 2011) where explaining away and other
causal-probabilistic relationships can be found highlights the importance of accurately
capturing these relationships in computational terms.

1.1.2 Bayesian belief networks: explaining evidence

Another important feature of BNs is that they can be used for both predictive rea-
soning and diagnostic reasoning (often referred to as an abduction, see, e.g., Korb and
4 Explanation in AI systems

Nicholson 2010). Consider the BN in Figure 1.1. An example of predictive reasoning

would be inferring a probability that the child has bruises given that it’s suffering
from haemophilia (i.e. inference from causes (h) to effects (e)); whereas diagnostic rea-
soning would be inferring a probability of physical abuse from learning that the child
has bruises (i.e. inference from effects (e) to causes (h)). Often, diagnostic reasoning
(abduction) is used to find the most probable explanations (causes) of observed evi-
dence (effects), that is to find the configuration h with the maximum p(h | e) (Pearl,
1988). Similarly, Shimony’s (1991) partial abduction approach first marginalizes out
variables that are not part of explanations
P (x) and then searches for the most probable
h: that is, find h with the maximum x p(h, x | e). More recently, Yuan et al. (2011)
introduced a method they call ‘Most Relevant Explanation’ (MRE) which chooses
the explanation that has the highest likelihood ratio compared to all other explana-
tions: that is, find h with the maximum p(e | h)/p(e | h), where h denotes all other
alternative explanations to h. Nielsen et al. (2008) introduced a ‘Causal Explanation
Tree’ (CET) method which uses the post-intervention distribution of variables (Pearl,
2000) in selecting explanations, which is in contrast to all previous methods since they
use non-interventional distribution of variables in a BN. Drawing on their definition of
causation, Halpern and Pearl (2005b) develop a definition of explanation to address a
question of why certain evidence holds given users epistemic state. Their definition of
explanation states that (i) a user should consider evidence to hold, (ii) an explanation
(h) is a sufficient cause of evidence, (iii) h is minimal (i.e. it does not contain irrelevant
or redundant elements), and (iv) h is not known at the beginning, but it is considered
as a possibility. This is an improvement to other accounts. However, their account of
causation has as an output again a set of variables in a BN model which are deemed as
causes of evidence in the model. Yap et al. (2008) employ Markov blanket to determine
which variables should feature in an explanation. A Markov blanket is of a node X
includes all nodes that are direct parents, children, or children’s parents of node X.
A powerful property of Markov blanket is that knowing the sates of all the variables
in a Markov blanket of X would uniquely determine the probability distribution of
X: additionally learning the states of other variables outside the Markov blanket of X
would not affect the probability distribution of X. Yap et al.’s ‘Explaining BN Infer-
ences’ procedure identifies Markov nodes of evidence (i.e. nodes in a Markov blanket
of the evidence node) and learns context specific independences in Markov nodes with
a decision tree to exclude irrelevant nodes in an explanation of the evidence.
Despite the difference among the methods, they all share at least one common-
ality: explanation of evidence is exhausted by a set of variables in a BN that these
methods have pointed to. In other words, evidence was provided a justification in terms
of a set of variables. This is undoubtedly useful, but in certain contexts (e.g. high-cost
domains, see Herlocker et al. 2000) it is arguably not enough to meet the demands
of user transparency. In contrast to the notion of explanation as a justification of ev-
idence is that of explanation of reasoning processes in BNs, and expert systems in
general (Lacave and Dı́ez, 2002; Sørmo et al., 2005; Wick and Thompson, 1992). Here
one is interested in how evidence propagates in a BN rather than in selecting a set of
variables that would account for evidence. Explaining reasoning processes in BNs been
a research focus amongst researchers for some time (see Lacave and Dı́ez 2002 for an
 Machine-generated explanation 5

overview). We next describe one more recent attempt in the context of the Bayesian
Argumentation via Delphi (BARD) project.

1.1.3 Bayesian belief networks: explaining reasoning processes

The BARD project (Cruz et al., 2020; Dewitt et al., 2018; Liefgreen et al., 2018;
Nicholson et al., 2020; Phillips et al., 2018; Pilditch et al., 2018; Pilditch et al., 2019)
set as its goal the development of assistive technology that could facilitate group
decision-making in an intelligence context. To this end, BARD provides a graphical
user interface enabling intelligence analysts to represent arguments as BNs and allow-
ing them to examine the impact of different pieces of evidence on arguments as well
as to bring groups of analysts to a consensus via an automated Delphi method. An
essential component of the system is the algorithm for generating natural language
explanations of inference in a BN, or more specifically, an explanation of evidence
propagation in a BN. This algorithm builds on earlier work by Zukerman and col-
leagues that have sought to use BNs to generate arguments (Zukerman et al., 1998;
Zukerman et al., 1999). The algorithm uses an evidence-to-goal approach to gener-
ate explanations for a BN. An explanation starts with the given pieces of evidence
and traces paths that describe their influence on intervening nodes until the goal is
reached. In essence, the algorithm adopts a causal interpretation of the links between
the connected nodes, finds a set of rules that describe causal relations in a BN, and
calculates all paths between evidence nodes and target nodes and builds corresponding
trees in order to determine the impact of evidence on target nodes. Figure 1.2 provides
an example. There we have four pieces of evidence: Emerson report, Quinns report,
and AitF Sawyer Report all stating that ‘The Spider’ is in the facility and Comms
Analyst Winter Report stating that ‘The Spider’ is not in the facility. The goal is to
explain the impact of these four pieces of evidence on two target variables, namely Is
The Spider in the facility? and Are logs true? (Are Emerson & Quinn spies?). The
algorithm first finds all relevant paths between evidence and target nodes, builds a
corresponding tree and calculates the impact of evidence on the target, which is sim-
ply a difference between the probability of the target node before learning particular
piece(s) of evidence and after learning particular piece(s) of evidence. This way the
algorithm can find HighImpSet—nodes that have the highest impact on the target
node, NoImpSet—nodes that, in light of the other evidence nodes, have no impact
on the target node, and OppImpSet—nodes that have the opposite impact to that of
HighImpSet. Finally, the algorithm realizes the explanations in English language using
sentences, clauses and phrases devised and combined by means of a semantic grammar
(Burton, 1976). The output of the algorithm is presented in Figure 1.3.
As can be seen, the output in Figure 1.3 provides significantly more information
to the user than just a single verdict on whether or not the variable Is The Spider
in the facility? is part of the explanation of the evidence, as would be the output
of methods looking for a justification of evidence. In addition to the impact sets, it
provides a natural language explanation on how different pieces of evidence influence
the probability of the target variable. Nevertheless, there remain continued challenges
with this approach. First, the algorithm retains difficulties in coping adequately with
soft evidence, namely evidence that we do not learn with probability 1. For instance,
6 Explanation in AI systems

Fig. 1.2 A BN of a fictional scenario used in BARD testing phase. Four pieces of evidence are
available: Emerson Report=Yes, Quinns Report=Yes, AitF Sawyer Report=Yes, and Comms
Analyst Winter Report=No.

Fig. 1.3 A summary report generated by the BARD algorithm applied on the BN from
Figure 1.2. In addition to the natural language explanation, it provides sets with nodes that
are HighImpSet, NoImpSet and OppImpSet. For the purposes of this chapter we can ignore
MinHIS and CombMinSet.

imagine that in the BN in Figure 1.2 we additionally learn that the logs are most
likely true, but we are not absolutely convinced. To reflect that we would set p(Are
logs true? (Are Emerson & Quinn spies) = True) to equal 0.95 for instance. Thus the
probability of Are logs true? (Are Emerson & Quinn spies) = True has changed from
0.46243 to 0.95, but it didn’t go all the way to 1. The current version of the algorithm
is not able to calculate the impact of such change. Second, the explanations generated
by the algorithm are not aimed specifically at what a human user might find hard to
understand. To make matters worse, it is arguably the interactions between variables
and the often counterintuitive effects of these, that users will most struggle with (for
psychological evidence to this effect see for example Dewitt et al. 2018; Liefgreen et al.
2018; Phillips et al. 2018; Pilditch et al. 2018; Pilditch et al. 2019; Tešić et al. 2020).
In other words, the system generates an (accurate) explanation, but not necessarily a
good explanation. For further guidance on what might count as a good explanation
we consult research on this topic within the philosophy of science and epistemology,
 Good explanation 7

where the topic has raised decades of interest.

1.2 Good explanation

1.2.1 A brief overview of models of explanation
The historic point of departure for thinking about the nature of explanation in phi-
losophy is the covering law model (Hempel and Oppenheim, 1948), also known as the
“deductive-nomological model” of scientific explanation (where nomological means
pertaining to the laws of nature). This model construes explanation as a deductive
argument with true premises that has the phenomenon to be explained (the so-called
explanandum) as its conclusion. Specifically, this conclusion is derived from general
laws and particular facts. For example, an explanation of a position of a planet at a
point in time consists of a derivation of that position from the Newtonian laws gov-
erning gravity (general law), and information about the mass of the sun, the mass of
the planet, and position at a particular time and velocity of each (particular facts)
(Woodward, 2017). A key feature of this model is that it views explanation and pre-
diction as essentially two sides of the same coin. In the same way that Newtonian
laws and information about the mass of the sun and the planet etc. can be used to
predict the position of the planet at some future time the inference can also be used
to explain the position of the planet after we observe it. In other words, we see here
the same tight coupling between diagnostic reasoning and predictive reasoning that
we mentioned earlier in the context of BNs. However, while this coupling works in
BN’s across the range of possible probabilities, it becomes forced in the covering law
model when dealing with probabilistic explanations, in particular, when dealing with
cases where the probability of observing the conclusion is low. Not only do probabilis-
tic contexts move the inference from deduction to an ampliative inference where the
conclusion is no longer certain, the symmetry between explanation and prediction also
becomes forced. We might for example readily explain someone being struck by the
lightning by appealing to stormy weather conditions and the fact that they were out
in the open. But we would nevertheless hasten to predict that someone will be struck
by the lightning even if they are out in the open and there is a storm as it is a low
probability event. This limits the utility of the covering law model within the social sci-
ences where deduction is not commonplace and where low probability events are often
found. Hempel himself was aware of these difficulties to the extent that he proposed
two versions of the model the deductive-nomological model and an inductive-statistical
one, and himself thought that the inductive statistical model applied only when the
explanatory theory involves high probabilities. Even this restriction, however, does
not deal appropriately with the asymmetries involved in explanation. These can be
observed even in purely deductive context as is illustrated by the following example
from Salmon (1992). Imagine there is a flagpole with a shadow of 20m and someone
asks why that shadow is 20m long. In this context, it seems appropriate to explain
the length of the shadow by appealing to the height of the flagpole, the position of
the sun, and the laws of trigonometry. These together adequately explain the shad-
ows length. But note that this inference can be reversed: when can also use the sun’s
position, the laws of trigonometry, and the length of the shadow to explain the height
of the flagpole. This, however, seems wrong; an adequate explanation of the height of
8 Explanation in AI systems

that flagpole presumably involves an appeal to the maker of the flagpole in some form
or other. Examples such as these serve to illustrate not just the limits of Hempel’s
account but of the limits of deductive approaches in the context of explanation more
generally.
The asymmetric relations involved in explanation prompted alternative accounts
of scientific explanation within the subsequent literature. Chief among these are causal
accounts which assert that to explain something is to give a specification of its causes.
The standard explication of cause in this context is that of factors without which
something could not be the case (i.e. conditio sine qua non). This deals readily even
with low probability events, and causes can be identified through a process of “screen-
ing off”. If one finds that p(M | N, L) = p(M | N ), then N screens off L from M
and that M is causally irrelevant to L. For example, a reading of a barometer (B)
and whether there is a storm (S) are correlated. However, knowing the atmospheric
pressure (A) will make these two independent: p(B | A, S) = p(B | A), suggesting no
causal relationship between B and S. However, the notion of cause in itself is notori-
ously fraught as is evidenced by J. L. Mackie’s convoluted (Mackie, 1965) definition
whereby a cause is defined as an “insufficient but necessary part of an unnecessary but
sufficient condition”. This rather tortured definition reflects the difficulties with the
notion of causation when multiple causes are present which is giving rise to overde-
termination (for example, decapitation and arsenic in the blood stream can both be
the causes of death), the difficulties created by causal chains (for example, tipping
over the bottle which hits the floor which releases the toxic liquid) and the impact
of background conditions (for example, putting yeast in the dough causes it to rise,
but only if it is actually put in the oven, the oven works, the electrical bills have been
paid, and so on). It is a matter of ongoing research to what extent causal Bayes nets,
that is BN’s supplemented with the do-calculus (Pearl, 2000), provide a fully satis-
factory account of causality and these difficulties (see also Halpern and Pearl 2005a).
At the same time, the difficulty of picking out a single one out of multiple potential
causes points to the second main alternative to Hempel’s covering law model, namely
so-called pragmatic accounts of explanation.
According to van Fraassen (1977) an explanation always has a pragmatic com-
ponent: specifically what counts as an explanation in any given context depends on
the possible contrasts the questioner has in mind. For example, consider the question
“why did the dog bury the bone?”. Different answers are required for different prosodic
contours: “why did the dog (i.e., not some other animal) bury the bone?”; why did
the dog bury the bone? (say, rather than eat it); why did the dog bury the bone? (say,
rather than the ball). In short, pragmatic accounts bring into the picture the recipient
of an explanation while rejecting a fundamental connection between explanation and
inference assumed by Hempel’s model.

1.2.2 Explanatory virtues

Philosophy has not only tried to characterise the nature of explanation, it has also
sought to identify the so-called “explanatory virtues”. Of the many things that might
count as an explanation according to a particular theoretical account of explanation,
not all may seem equally good or compelling. Among ‘explanations’, we might ask
 Good explanation 9

what distinguishes better ones from poorer ones. In search of explanatory virtues that
characterise good explanation, a number of factors have been identified: explanatory
power, unification, coherence, and simplicity are chief among these. Explanatory power
often relates to the ability of an explanation to decrease the degree to which we find the
explanandum surprising; the less surprising the explanadum in light of an explanation
the more powerful the explanation. For instance, a geologist may find a prehistoric
earthquake as explanatory of deformation in layers of bedrock to the extent that these
deformations would be less surprising given the occurrence of such an earthquake
(Schupbach and Sprenger, 2011). Unification refers to explanations’ ability to provide
a unified account of a wide range of phenomena. For example, Maxwell’s theory (expla-
nation) managed to unify electricity and magnetism (phenomena). Coherence renders
explanations that better fit our already established beliefs to be preferred to those
that do not (Thagard, 1989). Explanations can also have internal coherence, namely
how parts of an explanation fit together. An often motioned explanatory virtue is
simplicity. According to Thagard (1978), simplicity is related to the size and nature
of auxiliary assumptions needed by an explanation to explain evidence. For instance,
the phlogiston theory of combustion needed a number of auxiliary assumptions to ex-
plain facts that are easily explained by Lavoisier’s theory: it assumed existence of a
fire-like element ‘phlogiston’ that’s given away in combustion and that had ‘negative
weight’ since bodies undergoing combustion increase in weight. Others operationalise
simplicity as a number of causes invoked in an explanation: the more causes the less
simple an explanation (Lombrozo, 2007).
While all of these factors seem intuitive, debate persists about their normative
basis. In particular, there is ongoing debate within the philosophy of science about
whether these factors admit of adequate probabilistic reconstruction (Glymour, 2014).
At the same time, there is now a sizeable program within psychology that seeks to ex-
amine the application of these virtues to every day lay explanation. This body of work
probes the extent to which lay reasoners endorse these criteria when distinguishing
better from worse explanations (Bechlivanidis et al., 2017; Bonawitz and Lombrozo,
2012; Johnson et al., 2014a; Johnson et al., 2014b; Lombrozo, 2007; Lombrozo, 2016;
Pennington and Hastie, 1992; Sloman, 1994; Williams and Lombrozo, 2010; Zemla
et al., 2017). To date, researchers found some degree of support for these factors, but
also seeming deviations in practice.
Finally, there is a renewed interest in both philosophy and psychology in the no-
tion of inference to the best explanation (Harman, 1965; Lipton, 2003). Debate here
centres around the question of whether the fact that an explanation seems in some
purely non-evidential way better than its rivals should provide grounds for thinking
that explanation is more probable. In other words, the issue is whether an explanation
exhibiting certain explanatory considerations that other explanations do not should
be considered more likely to be true (Douven, 2013; Harman, 1967; Henderson, 2013;
Lipton, 2003; Thagard, 1978). Likewise this has prompted psychological research into
whether such probability boosts can actually be observed in reasoning contexts (Dou-
ven and Schupbach, 2015). The research on explanatory virtues in both philosophy
and psychology is still very much active.
10 Explanation in AI systems

1.2.3 Implications

What if anything can be inferred for the project of machine-generated explanation

from this body of research? First, the notion of explanation that emerges is a po-
tentially very different one across different parts of this literature. An explanation is
variously a hypothesis or a variable, and inference, or an answer to a question. From
afar, in the context of inference, we may distinguish explanation as a product from
explanation as a the process (Lombrozo, 2012). From a product perspective, an expla-
nation is a hypothesis or a claim that accounts for evidence when prompted to do so.
In contrast, explanation can also be viewed as a cognitive activity (process) that has
as its goal to generate explanation ‘products’. This distinction nicely corresponds to
what we have found in the literature on explanations in BNs reported in Section 1.1.
There, explanations are also viewed as products that consist of nodes in a BN that aim
to account for other nodes in a BN (i.e. evidence nodes) as well as reasoning processes
that include not just the final products but also how one arrives at these products.
This nicely illustrates how different disciplines can come to similar conceptual distinc-
tions without these distinctions being communicated from one to another. Establishing
the necessary communication channels will arguably help connect the research areas
working on closely related questions thus bringing different perspectives and inputs
into these areas.
This brings us to our second point. It seems clear that BNs provide a potential
tool that is compatible with present thinking about the explanation at least in prin-
ciple. They can capture the asymmetry in explanation as arcs are directed and can
have a causal interpretation (Pearl, 2000), whilst at the same time being able to make
predictions. This is in contrast to, for instance, a rule-based expert system with IF-
THEN rules and a set of facts which would be susceptible to the symmetry ‘error’ in
explanation illustrated by the flagpole example from Section 1.2.1. A BN on the other
hand would be able to account for the asymmetry given a causal interpretation and
directional representation of arrows. However, it is neither clear how explanations in
BNs can capture the pragmatic component that van Fraassen raises nor how to oper-
ationalise explanatory virtues in the context of BNs. These are all potential avenues
for further research.
Another point that can be drawn is that the debates about the nature of explana-
tion and explanatory virtues have been conducted at very high levels of abstraction.
They have also typically focused on philosophy of science and issues tightly related to
it. This is true even for psychological research on explanation, to the extent that it
has tried to model psychological investigations more or less directly on philosophical
distinctions. However, for the purposes of developing suitable AI algorithms, it does
also seems important to work in the opposite direction, as it were from the bottom up.
In other words, it seems important to simultaneously start with simple applications of
BN’s to multiple variable problems, and consider what kinds of explanations a human
(expert) would produce. This would shed light on the kinds of explanations that seem
natural and appropriate human users as well as provide guidelines on possible theories
of explanation. A similar point has been made in an AI literature with additionally
emphasising the importance of human-generate explanations serving as a baseline for
comparison with machine-generated explanations (Doshi-Velez and Kim, 2017). To
 Good explanation 11

explore these ideas further, we conducted a case study on explanation in BNs which
we describe next.

1.2.4 A brief case study on human-generated explanation

The main motivation of the study was to find out what kinds of explanations a human
(expert) would produce upon being presented with evidence in a BN. This is interesting
from both the psychological and the AI perspective as, on the one hand, it could give
us further inputs into human explanatory intuitions and preferences and, on the other
hand, it could inform the AI researcher that aims to build algorithms for an automated
generation of explanations.
In the study we used four BNs of different complexity found on a publicly avail-
able BN repository https://www.norsys.com/netlibrary/index.htm. The number
of nodes in the BNs ranged from 4 to 18 and the number of arcs raged from 4 to 20.
Figure 1.4 includes a BN used in the study.

Fig. 1.4 A BN used in the case study.

Three independent raters, all of whom were experts in probabilistic reasoning, were
then given access to implementations of the respective model in order to probe the
BNs in more detail, and asked to then provide answers to questions that prompted
them to consider how learning evidence changed the probabilities of the target nodes.
Below are sample questions:
• Given evidence: {Neighbours grass = wet}
Question: How does the probability of ‘Our Sprinkler = was on’ change compared
to when there was no evidence and why?
• Given Evidence: {Our grass = wet, Wall = wet}
Question: How does the probability of ‘Rain = rained’ change compared to when
the only available evidence was ‘Our Grass = wet’ and why?
Subsequently, the three independent sets of answers were subjected to an analysis
by a fourth person in order to identify both commonalities and differences across the
answers. This then formed the basis of the subsequent evaluation of those answers.
12 Explanation in AI systems

We describe the full set of results elsewhere (Tešić and Hahn, prepa), restricting
ourselves here to an initial summary of the results. First, we observed high levels of
agreement across answers. Differences were typically more presentational than sub-
stantive. For example, the following three statements all seek to describe the same
state of affairs:
• ‘As A is true C is more likely to be true if B is true and less likely to be true if B
is false. As we do not know B these alternatives essentially cancel themselves out
and leave the probability of C unchanged.’
• ‘It does not change. P (C | A) is equal to P (C) if P (C | A, B) = P (C | ∼A, ∼B)
and P (C | A, ∼B) = P (C | ∼A, B) (assuming P (B) = 0.5), which here is the case.’
• ‘According to model parameters: If A and B both true or both false, then C has
probability .75. If A true but B false, or vice-versa, then C has probability .25.
When we know A is true, and prior for B is 50%, there is a 50% that probability
of C is 75% and a 50% that probability of C is 2%, therefore overall probability
of C is 50%.’
Second, all appeal to hypothetical reasoning as a way of unpacking interactions of
evidence variables:
• ‘Wall = wet is a lot more likely if the sprinkler was on than if it rained (as a
matter of fact, if it rained, the wall is more likely to be dry than wet). Since, Our
sprinkler = was on went down, Wall = wet went down.’
Third, causal explanations are prevalent, and, where present, typically appeal to
the underlying target system being modelled (i.e., sprinkler, wall, rain) as opposed to
the model itself:
• ‘The probability of rain decreases because, although the sprinkler and rain can
both cause our grass to be wet, the wet wall is more likely to happen when the
sprinkler is on rather than rain.’
Notably, in appealing to causes, it is the most probable cause that seems to be
highlighted as an explanation:
• ‘There is a decrease [in probability] because the most likely cause of our grass
being wet is the sprinkler and since the wall is dry the sprinkler is unlikely to be
on.’
Finally, these data seem to suggest that the structure of the BNs is exploited
in order to zero in on ‘the explanation’ as a subset of all variables described in the
problem. Specifically, explanations seemed to make use of the Markov blanket: a set of
nodes consisting of the another node’s parents, its children, and its children’s parents
and making that node conditionally independent of the rest of the network (Korb and
Nicholson, 2010). In addition to Markov blanket, rater’s descriptions mostly followed
the direction of evidence propagation, i.e. followed the directed paths in a BN:
• ‘The probability of Battery voltage = dead increases because failure of the car to
start could be explained by the car not cranking and the likely cause of this is
a faulty starter system. A dead battery is one possible explanation for a faulty
starter system.’
 Bringing in the user: bi-directional relationships 13

This suggests that the explanatory virtue of ‘simplicity’ might, in a BN context,

be conceptualised in terms terms of a Markov blanket and path direction.
In summary, we see multiple features of the general philosophical literature reflected
in these explanations of everyday situations expressed with a BN model: a focus on
an inference or a reasoning process; the use of causal explanation for a probabilistic
system; the directional nature of explanation (its asymmetry); indications of pragmatic
sensitivity in that hypotheticals are used to express relevant ‘contrasts’; and, finally,
an emerging notion of simplicity in the use of the Markov blanket.
These results are, however, still very much preliminary further research is needed,
but hopefully they give us some sense of the kinds of explanations a human (expert)
may produce and, potentially, prefer.

1.3 Bringing in the user: bi-directional relationships

One of the themes of this chapter is that an explanation is an explanation for some-
one. We have already encountered this in pragmatic theories of explanation within the
philosophy literature, but it seems particularly pertinent to the AI context. As illus-
trated by the BARD assistive reasoning tool example, if explanations for AI systems
are to be effective, they must take the user into account. On the one hand, this will
require detailed research into what it is that the users of a given system do and do
not readily understand. These concerns will largely be specific to the context of the
AI system in question. But there are also further general considerations, which apply
across potential systems, that have not yet been adequately discussed.

1.3.1 Explanations are communicative acts

Only very recently has it been noted that the provision of an explanation, whether
from human or machine, is a communicative act. In order to understand the impact
of explanations, one must consequently consider the pragmatics of (natural language)
communication. Pragmatics, that is the part of language that deals with meaning in
context, tells us much about how users will come to interpret explanations. The need
for AI researchers to consider pragmatics in the context of machine-generated explana-
tion has recently been highlighted by Miller (2019) who very persuasively argues that
the research in explainable AI has largely neglected insights from social sciences, one
of which is a very important commonplace regarding the social aspect of explanations
where it is argued that explanations are often presented relative to the explainer’s
beliefs about the explainee’s beliefs. In particular, Miller argues that explanations go
beyond search for and presentation of associations and causes of evidence, but that
they are also contextual: explainer and explainee may have different background beliefs
regrading certain observed pieces of evidence and an explainable AI should address
this.
We briefly consider some further potential implications for the recipients of explana-
tions here. One general feature of communicative acts is that they provide information
about the speaker, intended or otherwise. Recent work concerned with the question of
trust has highlighted the interplay between culture and its contents and the perceived
reliability of the speaker (Olsson and Vallinder, 2013; Bovens and Hartmann, 2003).
One upshot of this, is that receiving an explanation is likely to change perceptions of
14 Explanation in AI systems

the reliability of the explanation’s source. Here, it is not only characteristics of the
explanation such as its perceived cogency, how articulately it is framed, or how easily
to process it that are likely to influence perceived source reliability, there are also likely
to be effects of the specific content. In particular, the extent to which the content of
the message fits with our present (uncertain) beliefs about the world has been shown
to affect beliefs about the issue at hand as well as beliefs about the perceived reliability
of the speaker (Collins et al., 2018; Collins and Hahn, 2019).

1.3.2 Explanations and trust

Because message content and source reliability/trustworthiness jointly determine the
impact of the communication on our beliefs, these interactions are likely to be con-
sequential for the extent to which the machine conclusion being explained is itself
perceived to be true. The literature in AI, in particular recommender systems, has
long recognized relationship between trust and explanation (Zhang and Chen, 2020).
Majority of research suggests that providing an explanation improves user’s trust in
an AI system (Herlocker et al., 2000; Sinha and Swearingen, 2002; Symeonidis et al.,
2009). However, the situation seems more intricate as more transparent systems do
not always lead to increase in trust (Cramer et al., 2008), and sometimes poor expla-
nations can lead to reduced acceptance of the AI systems (Herlocker et al., 2000). To
explore the interactions between explanations and trust, in addition to manipulation
transparency of AI systems, one would also need to experimentally manipulate a level
of trust users have in them. Recently, we have conducted empirical work on the rela-
tionship between reliability (which is related to the notion of trust as understood in
AI) and explanation in non-AI context (Tešić and Hahn, prepb). We have performed
three experiments where we used simple dialogues between two people (in the condi-
tion where an explanation was provided these were then explainer and explainee) on
five different issues to show that (i) providing an explanation for a claim increases not
just people’s convincingness in the claim but also their reliability of the person pro-
viding an explanation compared when there is no such explanation and (ii) providing
an explanation has a significantly greater impact on the convincingness and reliability
when people’s initial (prior) reliability of the source is low compared to when that
reliability is high. In the context of AI, these results suggest that providing a (good)
explanation of AI system’s decisions will arguably increase people’s convincingness
in/acceptance of these decisions as well as people’s perceived the reliability/trust of
the system. In particular, the impact of providing an explanation will be greater (and
most useful) if people’s initial perceived reliability/trust of an AI system is low.

1.3.3 Trust and fidelity

Recent surge of model agnostic post-hoc explanations of black-box deep learning mod-
els has significantly pushed the horizons of explainable AI, but at the same time it
has also introduced the fidelity problems. Namely, unlike explanations of BN where
original BN models could be used to generate explanations (either as justification
of evidence or as explanation of reasoning processes), deep learning models are not
transparent enough for either a lay or an expert human user to be able to explain
the models’ outputs; rather, one resorts to explanation models that are independent
 Conclusions 15

of deep learning models to generate explanations of these black-box models’ decisions

after these decisions have been made, i.e. post-hoc (Ribeiro et al., 2016; Zhang and
Chen, 2020). The explanations models are often model agnostic as they should be
able to explain decisions of any (black-box) model. Post-hoc model agnostic explana-
tion models have certainly furthered the work on explanation in AI, but they have
also prompted questions regarding the degree to which the explanations generated by
models are reflecting the real mechanisms that generated decisions of a deep learn-
ing model: i.e. they have raised questions regarding the fidelity of explanation models
(Sørmo et al., 2005; Ribeiro et al., 2016). In the literature, the trade-off between fi-
delity and interpretability of explanations models is often acknowledged: the higher
the fidelity of an explanation model to the black-box model the lower the interpretabil-
ity of that model and its transparency to a human user (Ribeiro et al., 2016). This
however brings trust into the consideration. On the one hand, if higher interpretability
is to increase trust, then trust may be negatively affected by higher fidelity. On the
other hand, if users expect higher fidelity explanation models, then lower fidelity may
now negatively affect trust. This potentially interesting relationship between fidelity
and trust is another open issue related to the interplay between a user and the system
that could be addressed in the future research.

1.3.4 Further research avenues

The effect of the communication on the reliability of the source/trust and possibly
fidelity, however, are unlikely to be the only way in which explanations alter beliefs
about what it is that is being explained. For example, does providing an explanation
constrain and/or make less ambiguous the underlying (causal) structure of the world
that the explainee had in mind before receiving the explanation (or in the case of a
BN, does providing an explanation restrict the number of potential BN structures that
the explainee has mind)? How providing an explanation of an (ab)normal event in a
causal chain of events reflects on our perceptions of that explanation? Does a detailed
explanation of a usual and obvious succession of events make that explanation less
preferred or worse compared to a less detailed explanation? All these questions call
for further investigation and can have implications for the explainable AI project.

1.4 Conclusions
We have seen multiple ways to build different notions of what counts as an explana-
tion. One of these involves explanation as identification of the variables that mattered
in generating certain outcome. In the context of computational models of explanation
in BN’s this corresponds to the usual focus on explaining observed evidence via unob-
served nodes within the network (Pacer et al., 2013). In other words, the explanation
identifies a justification/hypothesis. This is the notion of explanation that has figured
prominently in work on computer-generated explanations as well as in psychological
and philosophical literature on explanation. The second notion of explanation we con-
sidered includes explanation of the inference that links evidence and hypothesis. In
the context of BN’s this means explaining the inferences that lead to a change (or
no change) in the probabilities of the query nodes. In other words, the explanation
involves a target hypothesis plus information about the incremental reasoning process
16 Explanation in AI systems

that identifies that hypothesis. Finally, we considered the notion of explanation in

terms of providing understanding of what is hard to understand and or surprising to
the system’s user. Explanation in this widest sense requires not just identification of
a best hypothesis and explanation of the incremental steps that lead to the identifica-
tion of that hypothesis but also a user model that tells us what it is that human users
find difficult. For this widest sense of explanation, psychological research is essential.
Explanation so understood constitutes a fundamental problem of human computer
interaction and only empirical research that seeks to understand the human user can
lead to fully satisfactory answers.
Choosing among these three different notions of explanation, also directly affects
the answer to the question of what counts as a good explanation. As we saw in this
chapter, there is some guidance on the notion of good explanation that can be drawn
from both the philosophy and psychology literature; but it is also clear that more
specific work is required. In particular, most recent research suggests that such work
will need to take into account that providing an explanation is a communicative act
that changes perceptions of the communicator. In other words, explanation will not
merely translate an extant result into a language understood by the user, it will likely
affect how the user interprets the output of the system and the reliability of the system
itself. This means the provision of explanation will likely affect what the user considers
to be the verdict of the system in the first place, which could lead to further intricate
relationships between trust and concepts such as fidelity. It is thus essential that future
work on explanation within AI engage more fully with the pragmatic consequences of
communicating explanation.
References

Agrahari, Rupesh, Foroushani, Amir, Docking, T Roderick, Chang, Linda, Duns,

Gerben, Hudoba, Monika, Karsan, Aly, and Zare, Habil (2018). Applications of
Bayesian network models in predicting types of hematological malignancies. Scien-
tific reports, 8(1), 6951.
Bansal, Aayush, Farhadi, Ali, and Parikh, Devi (2014). Towards transparent systems:
Semantic characterization of failure modes. In European Conference on Computer
Vision, pp. 366–381. Springer.
Bechlivanidis, Christos, Lagnado, David A, Zemla, Jeffrey C, and Sloman, Steven
(2017). Concreteness and abstraction in everyday explanation. Psychonomic bulletin
& review , 24(5), 1451–1464.
Bonawitz, Elizabeth Baraff and Lombrozo, Tania (2012). Occam’s rattle: Children’s
use of simplicity and probability to constrain inference. Developmental psychol-
ogy, 48(4), 1156.
Bovens, Luc and Hartmann, Stephan (2003). Bayesian epistemology. Oxford Univer-
sity Press.
Burton, Richard R (1976). Semantic grammar: An engineering technique for con-
structing natural language understanding systems.
Chen, Jessie Y, Procci, Katelyn, Boyce, Michael, Wright, Julia, Garcia, Andre, and
Barnes, Michael (2014). Situation awareness-based agent transparency. Techni-
cal report, Army research lab Aberdeen proving ground MD human research and
engineering.
Chockalingam, Sabarathinam, Pieters, Wolter, Teixeira, André, and van Gelder,
Pieter (2017). Bayesian network models in cyber security: a systematic review.
In Nordic Conference on Secure IT Systems, pp. 105–122. Springer.
Choi, Arthur, Wang, Ruocheng, and Darwiche, Adnan (2019). On the relative ex-
pressiveness of Bayesian and neural networks. International Journal of Approximate
Reasoning, 113, 303–323.
Collins, PJ and Hahn, U (2019). We might be wrong, but we think that hedg-
ing doesn’t protect your reputation. Journal of experimental psychology. Learning,
memory, and cognition.
Collins, Peter J, Hahn, Ulrike, von Gerber, Ylva, and Olsson, Erik J (2018). The bi-
directional relationship between source characteristics and message content. Fron-
tiers in psychology, 9, 18.
Collobert, Ronan, Weston, Jason, Bottou, Léon, Karlen, Michael, Kavukcuoglu, Ko-
ray, and Kuksa, Pavel (2011). Natural language processing (almost) from scratch.
Journal of machine learning research, 12(Aug), 2493–2537.
Cramer, Henriette, Evers, Vanessa, Ramlal, Satyan, Van Someren, Maarten, Rut-
ledge, Lloyd, Stash, Natalia, Aroyo, Lora, and Wielinga, Bob (2008). The effects of
18 References

transparency on trust in and acceptance of a content-based art recommender. User

Modeling and User-Adapted Interaction, 18(5), 455.
Cruz, Nicole, Desai, Saoirse Connor, Dewitt, Stephen, Hahn, Ulrike, Lagnado, David,
Liefgreen, Alice, Phillips, Kirsty, Pilditch, Toby, and Tešić, Marko (2020). Widening
access to bayesian problem solving. Frontiers in Psychology, 11, 660.
Dardashti, Radin, Hartmann, Stephan, Thébault, Karim, and Winsberg, Eric (2019).
Hawking radiation and analogue experiments: A Bayesian analysis. Studies in His-
tory and Philosophy of Science Part B: Studies in History and Philosophy of Modern
Physics.
DARPA (2016). Explainable artificial intelligence (XAI) program. Retrieved from
https://www.darpa.mil/program/xplainable-artificial-intelligence.
Davis, Zachary and Rehder, Bob (2017). The causal sampler: A sampling approach
to causal representation, reasoning, and learning. In Proceedings of the Cognitive
Science Society.
Dewitt, Stephen, Lagnado, David A, and Fenton, Norman E (2018). Updating prior
beliefs based on ambiguous evidence. In Proceedings of the 40th Annual Conference
of the Cognitive Science Society.
Dizadji-Bahmani, Foad, Frigg, Roman, and Hartmann, Stephan (2011). Confirmation
and reduction: A Bayesian account. Synthese, 179(2), 321–338.
Doshi-Velez, Finale and Kim, Been (2017). Towards a rigorous science of interpretable
machine learning. arXiv preprint arXiv:1702.08608 .
Doshi-Velez, Finale, Kortz, Mason, Budish, Ryan, Bavitz, Chris, Gershman, Sam,
O’Brien, David, Schieber, Stuart, Waldo, James, Weinberger, David, and Wood,
Alexandra (2017). Accountability of ai under the law: The role of explanation.
arXiv preprint arXiv:1711.01134 .
Douven, Igor (2013). Inference to the best explanation, dutch books, and inaccuracy
minimisation. The Philosophical Quarterly, 63(252), 428–444.
Douven, Igor and Schupbach, Jonah N. (2015). The role of explanatory considerations
in updating. Cognition, 142, 299–311.
Drury, Brett, Valverde-Rebaza, Jorge, Moura, Maria-Fernanda, and de An-
drade Lopes, Alneu (2017). A survey of the applications of Bayesian networks
in agriculture. Engineering Applications of Artificial Intelligence, 65, 29–42.
Fallon, Corey K and Blaha, Leslie M (2018). Improving automation transparency:
Addressing some of machine learning’s unique challenges. In International Confer-
ence on Augmented Cognition, pp. 245–254. Springer.
Falzon, Lucia (2006). Using Bayesian network analysis to support centre of gravity
analysis in military planning. European Journal of operational research, 170(2),
629–643.
Felzmann, Heike, Villaronga, Eduard Fosch, Lutz, Christoph, and Tamò-Larrieux,
Aurelia (2019). Transparency you can trust: Transparency requirements for artificial
intelligence between legal norms and contextual concerns. Big Data & Society, 6(1),
2053951719860542.
Fenton, Norman, Neil, Martin, and Lagnado, David A (2013). A general structure for
legal arguments about evidence using Bayesian networks. Cognitive science, 37(1),
61–102.
 References 19

Fernbach, Philip M and Rehder, Bob (2013). Cognitive shortcuts in causal inference.
Argument & Computation, 4(1), 64–88.
Friedman, Nir, Geiger, Dan, and Goldszmidt, Moises (1997). Bayesian network clas-
sifiers. Machine learning, 29(2-3), 131–163.
Glymour, Clark (2014). Probability and the explanatory virtues. British Journal for
the Philosophy of Science, 66(3), 591–604.
Goodfellow, Ian, Bengio, Yoshua, and Courville, Aaron (2016). Deep Learning. MIT
Press.
Goodman, Bryce and Flaxman, Seth (2016). EU regulations on algorithmic decision-
making and a “right to explanation”. In ICML workshop on human interpretability
in machine learning (WHI 2016), New York, NY. http://arxiv. org/abs/1606.08813
v1.
Graves, Alex and Schmidhuber, Jürgen (2005). Framewise phoneme classifica-
tion with bidirectional lstm and other neural network architectures. Neural net-
works, 18(5-6), 602–610.
Grosan, Crina and Abraham, Ajith (2011). Rule-based expert systems. In Intelligent
Systems, pp. 149–185. Springer.
Gunning, David and Aha, David W (2019). Darpa’s explainable artificial intelligence
program. AI Magazine, 40(2), 44–58.
Hahn, Ulrike and Hornikx, Jos (2016). A normative framework for argument quality:
argumentation schemes with a Bayesian foundation. Synthese, 193(6), 1833–1873.
Hahn, Ulrike and Oaksford, Mike (2006). A Bayesian approach to informal argument
fallacies. Synthese, 152(2), 207–236.
Hahn, Ulrike and Oaksford, Mike (2007). The rationality of informal argumentation:
a Bayesian approach to reasoning fallacies. Psychological review , 114(3), 704.
Halpern, Joseph Y and Pearl, Judea (2005a). Causes and explanations: A structural-
model approach. Part I: Causes. The British journal for the philosophy of sci-
ence, 56(4), 843–887.
Halpern, Joseph Y and Pearl, Judea (2005b). Causes and explanations: A structural-
model approach. Part II: Explanations. The British journal for the philosophy of
science, 56(4), 889–911.
Harman, Gilbert (1965). The inference to the best explanation. The philosophical
review , 74(1), 88–95.
Harman, Gilbert (1967). Detachment, probability, and maximum likelihood. Nous,
401–411.
Harradon, Michael, Druce, Jeff, and Ruttenberg, Brian (2018). Causal learning and
explanation of deep neural networks via autoencoded activations. arXiv preprint
arXiv:1802.00541 .
Harris, Adam JL, Hahn, Ulrike, Madsen, Jens K, and Hsu, Anne S (2016). The
appeal to expert opinion: quantitative support for a Bayesian network approach.
Cognitive Science, 40(6), 1496–1533.
Hayes, Bradley and Shah, Julie A (2017). Improving robot controller transparency
through autonomous policy explanation. In 2017 12th ACM/IEEE International
Conference on Human-Robot Interaction (HRI, pp. 303–312. IEEE.
Hempel, Carl G. and Oppenheim, Paul (1948). Studies in the logic of explanation.
20 References

Philosophy of science, 15(2), 135–175.

Henderson, Leah (2013). Bayesianism and inference to the best explanation. The
British Journal for the Philosophy of Science, 65(4), 687–715.
Herlocker, Jonathan L, Konstan, Joseph A, and Riedl, John (2000). Explaining
collaborative filtering recommendations. In Proceedings of the 2000 ACM conference
on Computer supported cooperative work, pp. 241–250.
Howson, Colin and Urbach, Peter (2006). Scientific reasoning: the Bayesian approach.
Open Court Publishing.
Johnson, Samuel, Jin, Andy, and Keil, Frank (2014a). Simplicity and goodness-of-fit
in explanation: The case of intuitive curve-fitting. In Proceedings of the Annual
Meeting of the Cognitive Science Society, Volume 36.
Johnson, Samuel, Johnston, Angie, Toig, Amy, and Keil, Frank (2014b). Explanatory
scope informs causal strength inferences. In Proceedings of the Annual Meeting of
the Cognitive Science Society, Volume 36.
Korb, Kevin B. and Nicholson, Ann E. (2010). Bayesian artificial intelligence. CRC
press.
Krizhevsky, Alex, Sutskever, Ilya, and Hinton, Geoffrey E (2012). Imagenet classifi-
cation with deep convolutional neural networks. In Advances in neural information
processing systems, pp. 1097–1105.
Lacave, Carmen and Dı́ez, Francisco J (2002). A review of explanation methods for
Bayesian networks. The Knowledge Engineering Review , 17(2), 107–127.
Lagnado, David A, Fenton, Norman, and Neil, Martin (2013). Legal idioms: a frame-
work for evidential reasoning. Argument & Computation, 4(1), 46–63.
Laskey, Kathryn Blackmond and Mahoney, Suzanne M (1997). Network fragments:
Representing knowledge for constructing probabilistic models. In Proceedings of the
Thirteenth conference on Uncertainty in artificial intelligence, pp. 334–341. Morgan
Kaufmann Publishers Inc.
Liefgreen, Alice, Tešić, Marko, and Lagnado, David (2018). Explaining away: signif-
icance of priors, diagnostic reasoning, and structural complexity. In Proceedings of
the 40th Annual Conference of the Cognitive Science Society.
Lippmann, Richard, Ingols, Kyle, Scott, Chris, Piwowarski, Keith, Kratkiewicz,
Kendra, Artz, Mike, and Cunningham, Robert (2006). Validating and restoring
defense in depth using attack graphs. In MILCOM 2006-2006 IEEE Military Com-
munications conference, pp. 1–10. IEEE.
Lipton, Peter (2003). Inference to the best explanation. Routledge.
Lombrozo, Tania (2007). Simplicity and probability in causal explanation. Cognitive
psychology, 55(3), 232–257.
Lombrozo, Tania (2012). Explanation and abductive inference. Oxford handbook of
thinking and reasoning, 260–276.
Lombrozo, Tania (2016). Explanatory preferences shape learning and inference.
Trends in Cognitive Sciences, 20(10), 748–759.
Mackie, John L. (1965). Causes and conditions. American philosophical quar-
terly, 2(4), 245–264.
Madsen, Jens Koed, Hahn, Ulrike, and Pilditch, Toby D (2018). Partial source de-
pendence and reliability revision: the impact of shared backgrounds. In Proceedings
 References 21

of the 40th Annual Conference of the Cognitive Science Society.

Mercado, Joseph E, Rupp, Michael A, Chen, Jessie YC, Barnes, Michael J, Barber,
Daniel, and Procci, Katelyn (2016). Intelligent agent transparency in human–agent
teaming for multi-uxv management. Human factors, 58(3), 401–415.
Miller, Tim (2019). Explanation in artificial intelligence: Insights from the social
sciences. Artificial Intelligence, 267, 1–38.
Montavon, Grégoire, Samek, Wojciech, and Müller, Klaus-Robert (2018). Methods for
interpreting and understanding deep neural networks. Digital Signal Processing, 73,
1–15.
Morris, Michael W and Larrick, Richard P (1995). When one cause casts doubt on
another: A normative analysis of discounting in causal attribution. Psychological
Review , 102(2), 331–355.
Neapolitan, R. E. (2003). Learning Bayesian Networks. Upper Saddle River, NJ:
Prentice Hall.
Neil, Martin, Fenton, Norman et al. (2008). Using Bayesian networks to model the
operational risk to information technology infrastructure in financial institutions.
Journal of Financial Transformation, 22, 131–138.
Ng, Andrew Y and Jordan, Michael I (2002). On discriminative vs. generative clas-
sifiers: A comparison of logistic regression and naive bayes. In Advances in neural
information processing systems, pp. 841–848.
Nicholson, Ann E, Korb, Kevin B, Nyberg, Erik P, Wybrow, Michael, Zukerman,
Ingrid, Mascaro, Steven, Thakur, Shreshth, Alvandi, Abraham Oshni, Riley, Jeff,
Pearson, Ross et al. (2020). BARD: A structured technique for group elicitation of
bayesian networks to support analytic reasoning. arXiv preprint arXiv:2003.01207 .
Nielsen, Ulf, Pellet, Jean-Philippe, and Elisseeff, André (2008). Explanation trees for
causal Bayesian networks. Proceedings of the 24th Annual Conference on Uncer-
tainty in Artificial Intelligence (UAI-08), 427–434.
Olsson, Erik J and Vallinder, Aron (2013). Norms of assertion and communication
in social networks. Synthese, 190(13), 2557–2571.
Pacer, Michael, Williams, Joseph, Chen, Xi, Lombrozo, Tania, and Griffiths, Thomas
(2013). Evaluating computational models of explanation using human judgments.
In Proceedings of the 29th Conference on Uncertainty in Artificial Intelligenc.
Pearl, Judea (1988). Probabilistic reasoning in intelligent systems: Networks of plau-
sible inference. San Francisco, CA: Morgan Kauffman.
Pearl, Judea (2000). Causality: models, reasoning and inference. Volume 29. Springer.
Pennington, Nancy and Hastie, Reid (1992). Explaining the evidence: Tests of the
story model for juror decision making. Journal of personality and social psychol-
ogy, 62(2), 189.
Pernkopf, Franz and Bilmes, Jeff (2005). Discriminative versus generative parameter
and structure learning of Bayesian network classifiers. In Proceedings of the 22nd
international conference on Machine learning, pp. 657–664. ACM.
Pettigrew, Richard (2016). Accuracy and the Laws of Credence. Oxford University
Press.
Phillips, Kirsty, Hahn, Ulrike, and Pilditch, Toby D (2018). Evaluating testimony
from multiple witnesses: single cue satisficing or integration? In Proceedings of the
22 References

40th Annual Conference of the Cognitive Science Society.

Pilditch, Toby D., Fenton, Norman, and Lagnado, David (2019). The zero-sum fallacy
in evidence evaluation. Psychological Science, 30(2), 250–260.
Pilditch, Toby D, Hahn, Ulrike, and Lagnado, David A (2018). Integrating dependent
evidence: naı̈ve reasoning in the face of complexity. In Proceedings of the 40th Annual
Conference of the Cognitive Science Society.
Ramsey, Frank P (2016). Truth and probability. In Readings in Formal Epistemology,
pp. 21–45. Springer.
Rehder, Bob (2014). Independence and dependence in human causal reasoning. Cog-
nitive psychology, 72, 54–107.
Rehder, Bob and Waldmann, Michael R (2017). Failures of explaining away and
screening off in described versus experienced causal learning scenarios. Memory &
cognition, 45(2), 245–260.
Ribeiro, Marco Tulio, Singh, Sameer, and Guestrin, Carlos (2016). “Why should I
trust you?” Explaining the predictions of any classifier. In Proceedings of the 22nd
ACM SIGKDD international conference on knowledge discovery and data mining,
pp. 1135–1144.
Rieger, Laura, Chormai, Pattarawat, Montavon, Grégoire, Hansen, Lars Kai, and
Müller, Klaus-Robert (2018). Structuring neural networks for more explainable pre-
dictions. In Explainable and Interpretable Models in Computer Vision and Machine
Learning, pp. 115–131. Springer.
Rohekar, Raanan Y, Nisimov, Shami, Gurwicz, Yaniv, Koren, Guy, and Novik, Gal
(2018). Constructing deep neural networks by Bayesian network structure learning.
In Advances in Neural Information Processing Systems, pp. 3047–3058.
Roos, Teemu, Wettig, Hannes, Grünwald, Peter, Myllymäki, Petri, and Tirri, Henry
(2005). On discriminative Bayesian network classifiers and logistic regression. Ma-
chine Learning, 59(3), 267–296.
Rottman, Benjamin Margolin and Hastie, Reid (2014). Reasoning about causal rela-
tionships: Inferences on causal networks. Psychological bulletin, 140(1), 109–139.
Rottman, Benjamin M and Hastie, Reid (2016). Do people reason rationally about
causally related events? Markov violations, weak inferences, and failures of explain-
ing away. Cognitive psychology, 87, 88–134.
Salmon, Wesley C. (1992). Scientific explanation. In Introduction to the philosophy
of science (ed. M. H. Salmon, J. Earman, C. Glymour, J. Lennox, K. F. Schaffner,
W. C. Salmon, J. D. Norton, J. McGuire, P. Machamer, and J. G. Lennox), Chap-
ter 1, pp. 7–41. Englewood Cliffs, New Jersey, Prentice-Hall.
Samek, Wojciech, Wiegand, Thomas, and Müller, Klaus-Robert (2017). Explain-
able artificial intelligence: Understanding, visualizing and interpreting deep learning
models. arXiv preprint arXiv:1708.08296 .
Schupbach, Jonah N and Sprenger, Jan (2011). The logic of explanatory power.
Philosophy of Science, 78(1), 105–127.
Shimony, Solomon E (1991). Explanation, irrelevance and statistical independence.
In Proceedings of the ninth National conference on Artificial intelligence-Volume 1,
pp. 482–487. AAAI Press.
Sinha, Rashmi and Swearingen, Kirsten (2002). The role of transparency in rec-
 References 23

ommender systems. In CHI’02 extended abstracts on Human factors in computing

systems, pp. 830–831.
Sloman, Steven A (1994). When explanations compete: The role of explanatory
coherence on judgements of likelihood. Cognition, 52(1), 1–21.
Sørmo, Frode, Cassens, Jörg, and Aamodt, Agnar (2005). Explanation in case-based
reasoning–perspectives and goals. Artificial Intelligence Review , 24(2), 109–143.
Spiegler, Ran (2016). Bayesian networks and boundedly rational expectations. The
Quarterly Journal of Economics, 131(3), 1243–1290.
Sussman, Abigail B and Oppenheimer, Daniel M (2011). A causal model theory of
judgment. In Proceedings of the 33rd Annual Conference of the Cognitive Science
Society.
Symeonidis, Panagiotis, Nanopoulos, Alexandros, and Manolopoulos, Yannis (2009).
Moviexplain: a recommender system with explanations. In Proceedings of the third
ACM conference on Recommender systems, pp. 317–320.
Tešić, Marko (2019). Confirmation and the generalized Nagel–Schaffner model of
reduction: a Bayesian analysis. Synthese, 196(3), 1097–1129.
Tešić, Marko and Hahn, Ulrike (2019). Sequential diagnostic reasoning with indepen-
dent causes. In Proceedings of the 41th Annual Conference of the Cognitive Science
Society.
Tešić, Marko and Hahn, Ulrike (in prep.a). Human-generated explanations of infer-
ences in Bayesian networks: a case study.
Tešić, Marko and Hahn, Ulrike (in prep.b). The impact of explanation on explainee’s
beliefs and explainer’s perceived reliability.
Tešić, Marko, Liefgreen, Alice, and Lagnado, David (2020). The propensity inter-
pretation of probability and diagnostic split in explaining away. Cognitive Psychol-
ogy, 121, 101293.
Thagard, Paul (1989). Explanatory coherence. Behavioral and brain sciences, 12(3),
435–467.
Thagard, Paul R (1978). The best explanation: Criteria for theory choice. The journal
of philosophy, 75(2), 76–92.
Van Fraassen, Bas C. (1977). The pragmatics of explanation. American Philosophical
Quarterly, 14(2), 143–150.
Vineberg, Susan (2016). Dutch book arguments. In The Stanford Encyclopedia of
Philosophy (Spring 2016 edn) (ed. E. N. Zalta). Metaphysics Research Lab, Stanford
University.
Wachter, Sandra, Mittelstadt, Brent, and Floridi, Luciano (2017a). Why a right
to explanation of automated decision-making does not exist in the general data
protection regulation. International Data Privacy Law , 7(2), 76–99.
Wachter, Sandra, Mittelstadt, Brent, and Russell, Chris (2017b). Counterfactual
explanations without opening the black box: Automated decisions and the GDPR.
Harv. JL & Tech., 31, 841.
Wang, Hao and Yeung, Dit-Yan (2016). Towards Bayesian deep learning: A survey.
arXiv preprint arXiv:1604.01662 .
Wick, Michael R and Thompson, William B (1992). Reconstructive expert system
explanation. Artificial Intelligence, 54(1-2), 33–70.
 24 References

Wiegerinck, Wim, Burgers, Willem, and Kappen, Bert (2013). Bayesian networks,
introduction and practical applications. In Handbook on Neural Information Pro-
cessing, pp. 401–431. Springer.
Williams, Joseph J and Lombrozo, Tania (2010). The role of explanation in discovery
and generalization: Evidence from category learning. Cognitive Science, 34(5), 776–
806.
Woodward, James (2017). Scientific explanation. In The Stanford Encyclopedia of
Philosophy (Fall 2017 edn) (ed. E. N. Zalta). Metaphysics Research Lab, Stanford
University.
Xie, Peng, Li, Jason H, Ou, Xinming, Liu, Peng, and Levy, Renato (2010). Using
Bayesian networks for cyber security analysis. In 2010 IEEE/IFIP International
Conference on Dependable Systems & Networks (DSN), pp. 211–220. IEEE.
Yap, Ghim-Eng, Tan, Ah-Hwee, and Pang, Hwee-Hwa (2008). Explaining inferences
in bayesian networks. Applied Intelligence, 29(3), 263–278.
Yuan, Changhe, Lim, Heejin, and Lu, Tsai-Ching (2011). Most relevant explanation
in Bayesian networks. Journal of Artificial Intelligence Research, 42, 309–352.
Zemla, Jeffrey C, Sloman, Steven, Bechlivanidis, Christos, and Lagnado, David A
(2017). Evaluating everyday explanations. Psychonomic bulletin & review , 24(5),
1488–1500.
Zhang, Yongfeng and Chen, Xu (2020). Explainable recommendation: A survey and
new perspectives. Foundations and Trends in Information Retrieval , 14(1), 1–101.
Zukerman, Ingrid, McConachy, Richard, and Korb, Kevin B. (1998). Bayesian
reasoning in an abductive mechanism for argument generation and analysis. In
AAAI/IAAI, pp. 833–838.
Zukerman, Ingrid, McConachy, Richard, Korb, Kevin B., and Pickett, Deborah
(1999). Exploratory interaction with a Bayesian argumentation system. In IJCAI,
pp. 1294–1299.

View publication stats