Skip to main content
SpringerLink
Log in

Toward representing interpretation in factor-based models of precedent

  • Original Research
  • Published:
Artificial Intelligence and Law Aims and scope Submit manuscript

Abstract

This article discusses the desirability and feasibility of modeling precedents with multiple interpretations within factor-based models of precedential constraint. The main idea is that allowing multiple reasonable interpretations of cases and modeling precedential constraint as a function of what all reasonable interpretations compel may be advantageous. The article explains the potential benefits of extending the models in this way with a focus on incorporating a theory of vertical precedent in U.S. federal appellate courts. It also considers the costs of extending the models in this way, such as the significant increase in the functional size of the case base and the need to provide some kind of ordering on interpretations to select a “best” interpretation. Finally, the article suggests partially incorporating multiple interpretations of dimensions as a realistic starting point for incorporating interpretations generally, and shows how doing so can help address difficulties with dimensions. The conclusion remarks on the use of interpretations to deal with inconsistent precedents.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. We might add that the application of interpretation to facts occurs not only in decide the ultimate outcome of a case, but to decide each issue within the case, which then combine to produce the outcome. This would follow the four part approach found in (Bench-Capon 2023, sec 4; Collenette et al. 2023, sec. 2.1.4), where cases move from evidence to facts, facts to factors, factors to determinations of various issues, and then from those determinations to an final outcome.

  2. United States v Robinson 414 U.S. 218, 225 (1973). The example is discussed as part of an example of “narrowing” by Re (2016, p. 954).

  3. United States v. Wurie (2013), 728 F.3d, 3–7.

  4. See Riley v. California, 573 U.S. 403 (2014), in which the Supreme Court consolidated United States v. Wurie and Riley v. California. See discussion in Re (2016, p.955).

  5. This might matter a great deal. In the next section we will see that, if Broughton (2019) is correct, then an explanation in terms of distinguishing may characterize the court as having made an illegitimate decision while an explanation in terms of interpretation characterizes the court as having acted legitimately.

  6. The idea that U.S. judges will intentionally read past precedent in ways favorable to their preferred result is hardly new. It is one of the points emphasized (perhaps too much so) by the Legal Realists, such as Holmes (1897), and later by legal pragmatists (Posner 2008) and in social science approaches to the U.S. common law (Cross and Tiller 1998).

  7. He notes that the assumption of a best reading is controversial and subject to qualification (Re 2016, p. 928).

  8. State ex rel. Simmons v. Roper, 112 S.W.3d 397 (Mo. 2003).

  9. 492 U.S. 361 (1989).

  10. See also Simmons, 112 S.W.3d at 406, which comments on “flexible and dynamic” interpretation as a “fundamental premise” of Stanford.

  11. State ex rel. Simmons vs. Roper, 543 U.S. 551 (2005).

  12. But even here, notice the hedge between interpretation and application. “Implied exceptions” means the rule somehow contains the exceptions, but that’s not required for the rule to be defeasible.

  13. See Ashley and Brüninghaus ((2009), Al-abdulkarim et al. (2016a). Of course, this isn’t the only possible explanation.

  14. Under Broughton (2019)’s modifications R1 and R2 they are strict rules, not in virtue of the particular language used in the cases, but in virtue of their coming from a court higher up in the judicial hierarchy. That means any of the high court rules that are triggered and not trumped by other background rules will bind. Under Broughton’s modifications, R2 will bind here and the court must find for the defendant.

  15. E.g. (Horty 2011).

  16. (Rigoni 2015, 2018) are models of this kind. A different, but still permissive approach found can be found in the original HYPO (Rissland and Ashley 1987) as well as in more recent work (Bench-Capon and Atkinson 2022), which would be treat the case as introducing a dimension with respect to permissibility of different types of dogs. While (Rigoni 2018) represents dimensions as sets of factors, some of which are not present in the case, these other models represent dimensions as orderings of individual facts from which one can derive the presence of a factor. These models could represent the past court’s statement that “no dogs, not even service dogs, are allowed on buses” as establishing that the fact (rather than factor) that the dog is a service dog does not permit the derivation of a factor that favors the dog’s owner. I still consider these models permissive because they permit hypothetical facts (the past case doesn’t involve a service dog) to generate binding precedent.

  17. Re considers further kinds of rule modification when he discusses Scalia’s claim that the Supreme Court distinguished a previous case on an “accurate-in-fact (but inconsequential in principle distinction),” which sounds very much like distinguishing based on a difference in fact without any corresponding difference in reasons or factors (Re 2016, p. 935 (quoting Arizona State Legislature v. Arizona Independent Redistricting Committee, 576 U.S. 857 (2015) (Scalia, J., dissenting))). This would be illegitimate as distinguishing on any factor model, but Re says it would “qualify as legitimate narrowing” if it was based on a reasonable reading of the past precedent and violated no background principles. I do not follow Re on this point. Scalia seems to be criticizing the court for drawing such a distinction, not legitimating such behavior. He argues that if such a distinction was a legitimate basis for distinguishing, then the current case could also be so distinguished. However, he thinks the court is wrong to draw such distinctions. See Arizona State Legislature v. Arizona Independent Redistricting Committee, 576 U.S. 857-8 (2015) (Scalia, J., dissenting).

  18. See Janus v. American Federation of State, County, and Municipal Employees (2018), 138 S.Ct. 2472-4 (discussing special justifications for overruling).

  19. See Re (2014).

  20. Ignoring the purposes of models can cause much confusion. It leads to “toy” models being presented as accurate depictions, see Pfleiderer (2020).

  21. This approach is discussed in the context of analogical approaches, like (Stevens 2018), that allow for the current case to influence the interpretation of the past case directly in Rigoni (2021). In those analogical approaches, one past case may have different interpretations depending on the current case, so the casebase itself is relative to the current case. The approach here follows Re in treating interpretation as independent of the current case.

  22. It’s straightforward to extend this for reasoning with parts of precedents, holdings for different issues, etc.; I ignore that complexity here.

  23. See the work on automatically classifying cases according to factors in Ashley and Brüninghaus (2009), Branting et al. (2021), Gray et al. (2022, Mumford et al. (2022). I suspect it would be beneficial to break opinions up into issues and generate interpretations for each of those portions of the opinion, but that may raise problems of coherence—the combination of best interpretations for each issue may not be as coherent as other combinations of interpretations.

  24. A subtlety: Re reads “the best interpretation” subjectively, i.e. narrowing only happens when the judge or judges self-consciously apply a reading that they know is not the best (Re, 2019, n.37). Nonetheless, Re points to objective evidence to draw inferences about the judge’s subjective states. It is not clear that this is better than understanding “best interpretation” objectively, which would allow narrowing to capture instances of confirmation bias, wherein judges unconsciously adopt inferior readings because they support their favored outcome. Since the rule of the past case is effectively narrowed whether the judge knows it or not, the objective characterization seems superior in this context.

  25. In the same vein, Casey (2013, p.364) notes, “It also appears that in cases where an earlier opinion posits some holding h and a later opinion overrules the earlier opinion by positing ~ h, it may be possible to formulate a large number of hypothetical holdings that are implied by h (and may be extremely similar to h), but that are not contradicted by ~ h and thus are not deemed to have been overruled.”.

  26. The model assumes a pro-plaintiff or pro-defendant holding for each issue, and then labels the factors favoring a holding according to which party the holding favors. I ignore this complication.

  27. A similar example is given in Prakken (2021, p. 577).

  28. Dropping this assumption is critical to the Bench-Capon and Atkinson approach from 2017 onward (Bench-Capon and Atkinson 2017, 2021, 2022), which is discussed later in this section.

  29. “Entanglement” in a somewhat different sense was used in (Roth and Verheij 2004).

  30. The rest of this section makes use of a number of hypothetical cases. Please see Table 1 for a summary of each hypothetical case.

  31. Complications may arise from whether to treat the lack of information on percentage of income earned as a factor. For this example, I’m assuming the only factor in the current case is the 17 months stay.

  32. Prakken (2021) points out some similar issues with his approach when dealing with dimensions not mentioned in the ratio of the case. He notes that you can avoid those by essentially treating every dimension of the past case as part of the ratio and only allowing distinguishing in favor of side s if the value on some dimension in the current case is more favorable to s than the value on that dimension in the past case. That is, you can avoid the problem by using a results model approach with dimensions. However, that approach still will not work when the number of dimensions in each case varies.

  33. Many more options are available. Without constraint from background cases, the court can set a reference value of the factor in the ratio as lower than the value that occurs in the case. That is, the court could use \({M}_{d1, 12}^{def}\) in these ratios, because the value of 18 months would support the defendant as least as much as the value of 12 months.

  34. Case 1 using any one of Rule 1, Rule 2, or Rule 3 will force the result in Case 2, a feature discussed infra at Sect. 5.

  35. Horty (2021)’s theory can also construct Case 3 in such a way that it won’t bind Case 4. See infra at Sect. 5.

  36. This example might be taken to suggest that the extrema of dimensions be treated as factors, although it’s not clear that that changing from 100 to 95% of income earned abroad, for instance. As discussed infra at ** of the Bench-Capon and Atkinson approach addresses this issue in a slightly different way that enables it to treat certain values on a dimension as producing non-entangled factors.

  37. I’m thankful to an anonymous reviewer for explanation on this point.

  38. You could also represent the priority more strongly, with sufficient length being established as one of (Bruninghaus and Ashley 2003)’s “knock-out factors.”.

  39. Within the two-stage approach you could also model Case 5 as containing the entangled factor SufficientAbsenceGivenIncome, if you treat “unknown” as a value on the dimension for income earned. In Case 6 we would then have the factor InsufficientAbsenceGivenIncome unopposed. A number of complications arise from treating unknown values in this way, so I ignore it in the body text. But is an option available on the two stage views.

  40. Interestingly, on Horty’s (2021) approach, the Case 3 won’t bind Case 4 if we give Case 3 this ratio: \(\left\{{M}_{d1, 50}^{def}, {M}_{d2, 0\%}^{def}\right\}\to {\text{Defendant}}\). However, that is the “ratio” for that case under the results model as used by Oderkerken et al. (2023). Horty’s approach can get Case 3 to bind Case 4, as we saw, but it depends on the selection of the ratio. Case 3 must bind Case 4 on Oderkerken et al.’s view.

  41. See Carey (2013).

  42. See (Canavotto 2022, pp. 29–31). Her notion of support entails that c5 supports a pro-plaintiff ruling in c7 but c6 won’t support a pro-defendant ruling in c7.

References

  • Al-abdulkarim L, Atkinson K, Bench-Capon T (2016a) A methodology for designing systems to reason with legal cases using abstract dialectical frameworks. Artif Intell Law 24(1):1–49. https://doi.org/10.1007/s10506-016-9178-1

    Article  Google Scholar 

  • Al-abdulkarim L, Atkinson K, Bench-Capon TJM (2016b) Statement types in legal argument. In: Legal Knowledge and Information Systems–JURIX 2016: The Twenty-Ninth Annual Conference. IOS Press, (pp. 3–12)

  • Al-Abdulkarim L, Atkinson K, Bench-Capon TJM (2015) Factors, Issues and Values: Revisiting Reasoning with Cases. In: Proceedings of the 15th International Conference on Artificial Intelligence and Law. (pp. 3–12)

  • Ashley KD (1991) Reasoning with cases and hypotheticals in hypo. Int J Man-Mach Stud 34:753–796

    Article  Google Scholar 

  • Ashley KD, Brüninghaus S (2009) Automatically classifying case texts and predicting outcomes. Artif Intell Law 17(2):125–165. https://doi.org/10.1007/s10506-009-9077-9

    Article  Google Scholar 

  • Atkinson K, Bench-Capon TJM, Prakken H, Wyner A (2013) Argumentation schemes for reasoning about factors with dimensions. In: Ashley KD (ed) Proceedings of 26th International Conference on Legal Knowledge and Information Systems (JURIX 2013). IOS Press, Amsterdam, (pp. 39–48)

  • Bench-Capon T (1999) Some observations on modelling case based reasoning with formal argument models. In: Proceedings of the 7h International Conference on Artificial intelligence and Law. pp 36–42

  • Bench-Capon T (2023) Whatever happened to hypotheticals? In: Proceedings of the 19th International Conference on Artificial Intelligence and Law. Association for Computing Machinery, (pp. 387–391)

  • Bench-Capon T, Atkinson K (2022) Argument schemes for factor ascription. In: Toni F, Polberg S, Booth R, et al. (eds) Proceedings of COMMA. IOS Press, (pp. 68–79)

  • Bench-Capon T, Atkinson K (2021) Precedential constraint: the role of issues. Association for Computing Machinery

    Book  Google Scholar 

  • Bench-Capon T, Atkinson K (2017) Dimensions and values for legal CBR. In: Wyner A, Casini G (eds) Legal knowledge and information systems. IOS Press, pp 27–32

    Google Scholar 

  • Branting LK (2020) Explanation in hybrid, two-stage models of legal prediction. In: XAILA Workshop at JURIX. (pp. 1–10)

  • Branting LK, Pfeifer C, Brown B et al (2021) Scalable and explainable legal prediction. Artif Intell Law 29(2):213–238. https://doi.org/10.1007/s10506-020-09273-1

    Article  Google Scholar 

  • Broughton GL (2019) Vertical precedents in formal models of precedential constraint. Artif Intell Law 27(3):253–307. https://doi.org/10.1007/s10506-019-09244-1

    Article  Google Scholar 

  • Bruninghaus S, Ashley K (2003) Predicting outcomes of case based legal arguments. In: Proceedings of 9th International Confernece on Artificial Intelligence Law, (pp. 233–42)

  • Canavotto I (2022) Precedential constraint derived from inconsistent case bases. Front Artif Intell Appl 362:23–32. https://doi.org/10.3233/FAIA220445

    Article  Google Scholar 

  • Carey M (2013) Holdings about holdings: Modeling contradictions in judicial precedent. Artif Intell Law 21(3):341–365. https://doi.org/10.1007/s10506-013-9141-3

    Article  Google Scholar 

  • Collenette J, Atkinson K, Bench-Capon T (2023) Explainable AI tools for legal reasoning about cases: a study on the European court of human rights. Artif Intell 317:103861. https://doi.org/10.1016/j.artint.2023.103861

    Article  MathSciNet  Google Scholar 

  • Cross FB, Tiller EH (1998) Judicial partisanship and obedience to legal doctrine: Whistleblowing on the Federal Courts of Appeals, Yale Law J. 107

  • Dworkin R (1986) Law’s empire. Harvard University Press, Cambridge

    Google Scholar 

  • Gray M, Savelka J, Oliver W, Ashley KD (2022) Toward automatically identifying legally relevant factors. In: Francesconi E et al (eds) Legal knowledge and information systems. IOS Press, pp 53–62

    Google Scholar 

  • Horty J (2021) Modifying the reason model. Artif Intell Law 29(2):271–285. https://doi.org/10.1007/s10506-020-09275-z

    Article  Google Scholar 

  • Horty JF (2017) Reasoning with dimensions and magnitudes. In: Proceedings of the 15th International Conference on Artificial Intelligence and Law. ACM Press, (pp. 109–118)

  • Horty JF (2015) Constraint and freedom in the common law. Philos Impr 15(25):1–27

    Google Scholar 

  • Horty JF (2019) Reasoning with dimensions and magnitudes. Artif Intell Law 27(3):309–345

    Article  Google Scholar 

  • Horty JF (2011) Rules and reasons in the theory of precedent. Leg Theory 17(01):1–33

    Article  Google Scholar 

  • Horty JF, Bench-Capon TJM (2012) A factor-based definition of precedential constraint. Artif Intell Law 20(2):181–214

    Article  Google Scholar 

  • Mullins R (2022) Two factor-based models of precedential constraint: a comparison and proposal. Artif Intell Law. https://doi.org/10.1007/s10506-022-09335-6

    Article  Google Scholar 

  • Mumford J, Atkinson K, Bench-Capon T (2022) Reasoning with legal cases: a hybrid ADF-ML approach. Legal knowledge and information systems. IOS Press, pp 93–102

    Google Scholar 

  • Odekerken D, Bex F, Prakken H (2023) Justification, stability and relevance for case-based reasoning with incomplete focus cases. In: Proceedings of the 19th International Conference on Artificial Intelligence and Law. (pp. 177–186)

  • Pfleiderer P (2020) Chameleons: the misuse of theoretical models in finance and economics. Economica 87(345):81–107. https://doi.org/10.1111/ECCA.12295

    Article  Google Scholar 

  • Posner RA (2008) How judges think. Harvard University Press, Cambridge

    Google Scholar 

  • Prakken H (2021) A formal analysis of some factor- and precedent-based accounts of precedential constraint. Artif Intell Law 29(4):559–585

    Article  Google Scholar 

  • Raz J (2009) The authority of law : essays on law and morality. Oxford University Press, Oxford

    Google Scholar 

  • Re R (2016) Narrowing supreme court precedent from below. Georgetown Law J 104(1):921–971

    Google Scholar 

  • Re R (2014) Narrowing precedent in the supreme court. Columbia Law Rev 114:1861–1911

    Google Scholar 

  • Rigoni A (2015) An improved factor based approach to precedential constraint. Artif Intell Law 23(2):133–160. https://doi.org/10.1007/s10506-015-9166-x

    Article  Google Scholar 

  • Rigoni A (2021) AA-RM wrestling: comparing analogical approaches and rule models for legal reasoning. Leg Theory 27(3):207–235. https://doi.org/10.1017/S135232522100015X

    Article  Google Scholar 

  • Rigoni A (2018) Representing dimensions within the reason model of precedent. Artif Intell Law 26(1):1–22. https://doi.org/10.1007/s10506-017-9216-7

    Article  MathSciNet  Google Scholar 

  • Rissland E, Ashley K (1987) HYPO: a precedent-based legal reasoner. Amherst, MA

  • Roth B, Verheij B (2004) Dialectical arguments and case comparison. Legal knowledge and information systems. Jurix 99–108

  • Simpson AWB (1961) The ratio decidendi of a case and the doctrine of binding precedent. In: Guest AG (ed) Oxford essys in jurisprudence: a collaborative work. Oxford University Press, London, pp 148–175

    Google Scholar 

  • Stevens K (2018) Reasoning by precedent– between rules and analogies. Leg Theory 24(3):216–254. https://doi.org/10.1017/S1352325218000113

    Article  Google Scholar 

  • Wendell Holmes O (1897) The path of the law. Harv Law Rev 10:457

    Article  Google Scholar 

Cases Cited

  • Arizona State Legislature v. Arizona Independent Redistricting Commission et al

  • Arizona Independent Redistricting Commission, 576 U.S. 787 (2015)

  • Janus v. American Federation of State, County, and Municipal Employees, Council 31, 585 U.S. __ (2018) [abbreviated Janus v. AFSCME]

  • Riley v. California, 573 U.S. 373 (2014)

  • Stanford v. Kentucky, 492 U.S. 361 (1989)

  • State Ex rel Simmons v. Roper, 112 S.W. 3d 397 (Mo. 2003)

  • State Ex rel Simmons v. Roper, 543 U.S. 551 (2005)

  • United States v. Robinson, 414 U.S. 218 (1973)

  • United States v. Wurie, 728 F.3d 1 (1st Cir. 2013)

Download references

Author information

Authors and Affiliations

  1. Barrett Honors College, Arizona State University, Tempe, AZ, USA

    Adam Rigoni

Authors
  1. Adam Rigoni
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Adam Rigoni.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Rigoni, A. Toward representing interpretation in factor-based models of precedent. Artif Intell Law (2024). https://doi.org/10.1007/s10506-023-09384-5

Download citation

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s10506-023-09384-5

Keywords

Search

Navigation