Snap Shot
- The question: Does describing a possibility in detail, broken into its specific parts, make it seem more probable or more frequent than describing the same possibility as a single whole, and does that matter for witnesses, jurors, lawyers, and experts?
- What the research shows: Judged probabilities of the separate parts of an event often add up to more than the judged probability of the whole, and possibilities left unmentioned tend to be underweighted. The pattern appears in physicians’ and lawyers’ judgments, in people’s estimates of how often they did things, and in mock jurors’ verdicts when a specific alternative suspect is described.
- What it does not show: Unpacking does not always raise judgments; with typical components it may have no effect, and with atypical components it can lower them. Most of the research concerns judgment rather than memory, and none of it can say whether a particular estimate, alternative account, or charge is correct.
- The law: The prosecution must prove guilt beyond a reasonable doubt (In re Winship), and Arizona’s instruction asks whether there is a “real possibility” the defendant is not guilty (Portillo). Expert testimony on probability judgment is governed by Rule 702 and Houser, and in Arizona by Chapple and Nordstrom, without opinions on a witness’s credibility.
- Issues for each side: Both sides present possibilities in detail, so the effect favors neither; prosecutors should ask whether an alternative account rests on evidence or on vivid itemization, and defense counsel should present alternatives the evidence supports rather than multiplying weak ones.
Most of the papers in this hub concern memory. This one concerns judgment: how people estimate how probable or how frequent something is. The subadditivity effect refers to a common pattern in which the probabilities people give to the separate parts of a category add up to more than the probability they give to the category as a whole. Seen from the other direction, breaking a possibility into specific components, a process researchers call unpacking, tends to make it seem more likely than describing it as one undivided whole, even though the underlying events are identical.
The effect reaches court because legal fact-finding is full of judgments about likelihood. Jurors weigh whether an alternative explanation is a real possibility, lawyers forecast trial outcomes and advise on settlement, experts testify about the probability of a diagnosis or an outcome, and witnesses are asked how often something happened. Each judgment can depend on how the possibilities are described.
This paper explains support theory, the leading account of the effect; the evidence on unpacking in probability and frequency estimates; the conditions under which unpacking fails or reverses; and the research on jurors, lawyers, and experts. It states candidly how limited the connection to memory is, then turns to the limits of the evidence, the law, and the issues for each side.
Support Theory
Tversky and Koehler (1994) proposed that subjective probability attaches not to events themselves but to descriptions of events. A description calls to mind evidence, or support, for the possibility it names. A detailed description that lists specific components tends to bring more support to mind than a compact one, so it receives a higher probability. One illustration in that work concerned causes of death: the probabilities respondents assigned separately to death from heart disease, cancer, and other natural causes summed to more than the probability other respondents assigned to death from natural causes as a whole, although the categories describe the same set of deaths. Judgments with this property, in which the whole receives less than its components combined, are called subadditive.
Rottenstreich and Tversky (1997) extended the theory. They distinguished implicit subadditivity, in which an unpacked description of an event receives a higher probability than the packed description, from explicit subadditivity, in which the probability of a list of possibilities judged together is no more than the sum of the probabilities of its components judged one at a time. In several studies of both probability and frequency judgment they found both patterns, attributing the first to the greater availability of the unpacked components and the second to people mentally repacking a list and anchoring on part of it.
A related and older finding concerns what is left out. Fischhoff et al. (1978) showed people fault trees, diagrams listing the reasons a car might fail to start. When branches were removed, people did not assign the missing causes to the remaining “all other problems” category as they should have; they treated the shortened list as nearly complete. Experienced mechanics showed the same tendency. Possibilities that are explicitly listed tend to loom larger, and those that are not listed tend to be underweighted.
Where Unpacking Fails or Reverses
Unpacking is not a lever that always moves judgments upward. Sloman et al. (2004) compared judgments of packed descriptions, such as the probability that a businessman does business with a European country, with unpacked descriptions that named specific instances. When the named instances were typical, unpacking had no effect; when they were atypical, unpacking lowered the judged probability of the whole, a pattern called superadditivity. The authors concluded that support theory, as formulated, could not account for these results.
More recent work has challenged another assumption of the theory, that a given description always brings the same amount of support to mind. In three studies in which participants estimated probabilities from relevant statistics, the judged probability of a target possibility depended on the other possibilities under consideration: it rose when an obviously weaker alternative was included and fell when a similar alternative was included (Cai & Pleskac, 2023). Judgments of likelihood are therefore shaped not only by how a possibility is described but by what it is compared against.
Unpacking also extends beyond probability. In five preregistered experiments with 3,335 adults, describing a group in terms of its constituent subgroups shifted judgments of blame toward that group across three different intergroup conflicts, among both partisans and nonpartisans (Halevy et al., 2022). The study did not involve legal cases, but it suggests that how a set of potential wrongdoers is described can affect where people place responsibility.
Frequency Estimates and the Connection to Memory
The effect is primarily a phenomenon of judgment, and its relevance to memory is limited. The clearest link concerns estimates of how often something happened. Mulford and Dawes (1999) found that people’s estimates of how frequently they had engaged in personal activities were subadditive: estimates for specific components summed to more than estimates for the whole category, even when the components made up only part of the category. The authors noted that the level of detail at which a question is asked can produce quite different reports of the same events. Rottenstreich and Tversky (1997) likewise reported subadditivity in frequency judgments.
Duration estimates show a related pattern. In studies of a recorded mock crime, witnesses’ estimates for each component of the event did not sum to their estimate for the whole, and the sum of the parts was less biased than the overall estimate (Gasper et al., 2019). In that research, then, breaking an event into parts improved the duration estimate rather than distorting it, which is a reminder that unpacking is not inherently a source of error.
These findings bear on questions witnesses are often asked, such as how many times an event occurred or how long a series of events lasted. It is a reasonable inference, though not one tested directly in forensic settings, that a count obtained by asking about each place, period, or type of incident separately may differ from a count obtained by asking about all of them at once, and that neither figure is necessarily more accurate. The research does not show that a witness who gives different counts in response to differently structured questions is unreliable or untruthful; it shows that frequency estimates are judgments built partly from the question, which is why the form of the question should be recorded. The related influence of a suggested number on estimates is discussed in the paper on anchoring in this hub.
Probability Judgments by Jurors, Lawyers, and Experts
The most direct legal study concerned reasonable doubt. In a study of 253 undergraduate mock jurors (the number per condition was not located), participants read the transcript of a circumstantial murder case in which the defense either rebutted the prosecution’s evidence and argued reasonable doubt or, in addition, identified one, two, or three specific alternative suspects, each supported by some circumstantial evidence. Naming one alternative suspect reduced guilty verdicts from about 73% to about 35%, adding a second or third suspect produced no significant further reduction, and participants still rated the defendant as more likely guilty than any single alternative suspect (Tenney et al., 2009). The study shows that a concrete alternative can change how mock jurors apply the standard of proof; it was a written mock trial without deliberation, each alternative suspect was supported by some circumstantial evidence, and it does not show that any particular alternative is true or false.
Lawyers show the same pattern in forecasting. In six studies of tort, child custody, and antitrust scenarios, practicing attorneys’ probability estimates for complementary outcomes, and for outcomes broken into components, tended to sum to more than one, and the way outcomes were described was associated with the attorneys’ recommendations about settlement (Fox & Birke, 2002). Physicians given brief clinical scenarios likewise discounted diagnostic and prognostic possibilities that were not specifically named (Redelmeier et al., 1995).
These findings matter for expert testimony. An expert asked to estimate a probability intuitively, such as the likelihood of a medical cause, an accident mechanism, or a diagnostic outcome, may give a different answer depending on whether the alternatives are listed or summarized as “other causes,” and the same is true of the lawyer who frames the question. Probability estimates drawn from data, such as base rates in a validated study, are less exposed to this effect than intuitive estimates, although the way such figures are described to a factfinder can still affect how they are understood. Because expertise did not eliminate the effect among physicians, lawyers, or mechanics, an expert’s experience is not by itself a safeguard.
Limits of the Evidence
Most subadditivity research uses student or online participants, hypothetical scenarios, and general questions with no personal consequences. The juror study used a written transcript, undergraduates, and individual verdicts without deliberation, and it has not been widely replicated with jury-eligible samples or deliberating groups. The size and even the direction of unpacking effects depend on the typicality and plausibility of the components (Sloman et al., 2004) and on the alternatives being compared (Cai & Pleskac, 2023). Some causes-of-death and similar illustrations come from studies more than thirty years old. The memory research is thin: a small number of studies show subadditivity in frequency and duration estimates, and none tests how question structure affects actual witnesses’ counts of repeated events. Group research cannot establish that a particular juror, lawyer, expert, or witness was influenced by how a possibility was described, and it cannot say which description produces the more accurate judgment in a given case.
Questions to Ask About This Research
When the causes-of-death illustration from Tversky and Koehler (1994) is cited, counsel should ask what was actually judged: general probability estimates by research participants, not verdicts or diagnoses about a specific event. When Tenney et al. (2009) is offered to show that a concrete alternative-suspect account raises reasonable doubt, counsel should ask who was studied and whether the conclusion follows for this case: undergraduates reading a transcript are not deliberating jurors hearing live evidence, the alternatives were supported by some evidence, and additional suspects beyond the first added no significant effect. When unpacking research is used to suggest that a defense alternative appears strong only because it is itemized, counsel should ask whether the same argument applies to an elaborately subdivided prosecution theory, and whether the listed components are representative of the category, since unusual components have lowered rather than raised judged probability (Sloman et al., 2004). When an expert offers a probability, counsel should ask whether it was derived from data or estimated intuitively, and how the alternatives were described when the estimate was formed (Fox & Birke, 2002; Redelmeier et al., 1995). These questions are explained in “Reading Research Claims: A Guide for Attorneys” on this site.
The Subadditivity Effect and the Law
Federal and Military Standards
The Due Process Clause requires proof beyond a reasonable doubt of every fact necessary to constitute the crime charged (In re Winship, 1970), and the Constitution does not require any particular form of words in defining reasonable doubt (Victor v. Nebraska, 1994). The subadditivity research bears on how factfinders evaluate alternative possibilities under that standard, but it does not change the standard or tell a factfinder how much weight an alternative deserves. Expert testimony about probability judgment is governed by Federal Rule of Evidence 702, as amended December 1, 2023, under which the proponent must show the court that it is more likely than not that the rule’s requirements are met, including that the opinion reflects a reliable application of reliable methods to the case. Daubert v. Merrell Dow Pharmaceuticals, Inc. (1993) made the trial judge a gatekeeper, and Kumho Tire Co. v. Carmichael (1999) extended that obligation to all expert testimony, including testimony based on technical or other specialized knowledge. In courts-martial, Military Rule of Evidence 702 tracks the federal rule, the 2023 federal amendment took effect in the military rule in 2025 through Military Rule of Evidence 1102 (Joint Service Committee on Military Justice, 2025), and practitioners should confirm current text; military judges apply the six factors of United States v. Houser (1993). Such testimony is general and educational; no reported decision admitting or excluding it on this topic was located, and it cannot tell the factfinder whether a particular alternative is credible.
Arizona
Arizona trial courts must give the reasonable doubt instruction approved in State v. Portillo (1995) in every criminal case; it appears in the Revised Arizona Jury Instructions (Criminal) as Standard Criminal 4(a) (Burden of Proof) and tells jurors that they must find the defendant guilty if they are firmly convinced of guilt and must give the defendant the benefit of the doubt if they think there is a real possibility that the defendant is not guilty. Arizona Rule of Evidence 702 adopted the federal formulation and the Daubert standard effective January 1, 2012, and was amended effective January 1, 2024, to conform to the 2023 federal amendment (Ariz. Sup. Ct. Order R-23-0004). State v. Chapple (1983) is Arizona’s leading case on expert testimony about memory and eyewitness reliability; excluding such testimony was an abuse of discretion on its facts, while admission otherwise remains within the trial court’s discretion (State v. Nordstrom, 2001). An expert may describe general principles but may not opine on whether a particular witness is truthful (State v. Lindsey, 1986, 149 Ariz. at 475).
Special Issues in Legal Practice
The principles described above apply with equal force to every party in a legal proceeding. The science operates the same way regardless of who is asking the questions, and the research summarized here does not by itself favor either side. What differs between the parties is not the science but the vantage point — the particular places where these issues tend to surface, and the particular errors each party is best positioned to guard against. The two summaries below identify those recurring issues. They are offered for orientation, not as litigation strategy; in every case the controlling question is what the evidence in that case will, and will not, support, and the expert’s role is to inform the trier of fact rather than to advocate for either side.
Special Issues for Prosecutors
The research supports several propositions prosecutors can legitimately present. The apparent strength of an alternative account can depend on how concretely it is described rather than on the evidence for it, and in the mock-juror research the defendant was still judged more likely guilty than any single alternative suspect. Adding further alternatives beyond the first did not add significantly to doubt, and atypical or implausible components can lower rather than raise the judged probability of an account. Prosecutors are well positioned to ask the factfinder to evaluate each alternative by the evidence that supports it, and to ensure that witness counts and frequency estimates are obtained with open questions and recorded in the form in which they were asked.
The research also counsels caution. The same description-dependence works on the prosecution’s theory: a case presented as a detailed sequence of specific acts or a list of corroborating circumstances can feel more probable than its evidence warrants. A difference between a witness’s count given in answer to a general question and a count built up from specific questions is an expected feature of frequency judgment, not by itself evidence of either accuracy or inaccuracy. Characterizing a supported alternative as merely a framing device goes beyond what the research shows.
Special Issues for Defense
The research supports several propositions defense counsel can legitimately present. A concrete alternative account supported by evidence can properly bear on whether a reasonable doubt exists, and possibilities that are not stated tend to be underweighted. Physicians, lawyers, and other experts show subadditive judgments, so an expert’s intuitive probability estimate can be tested by asking how the alternatives were framed, and a witness’s frequency estimate can be tested by asking how the question that produced it was structured.
The research also counsels caution about its limits. Unpacking does not reliably increase judged probability, and implausible or multiplied alternatives may add nothing to doubt. The mock-juror findings do not establish that any particular alternative is true, and the same effect applies to defense experts and to defense presentations. Differences between counts obtained through different questions do not show that a complainant or witness is mistaken or untruthful, and an expert cannot tell the factfinder which figure is correct.
This paper is for educational purposes only and is not legal advice. The author is a psychologist, not an attorney; summaries of law are general and must be verified against current authority or with a licensed attorney.
References
Cai, X., & Pleskac, T. J. (2023). When alternative hypotheses shape your beliefs: Context effects in probability judgments. Cognition, 231, Article 105306. https://doi.org/10.1016/j.cognition.2022.105306
Fischhoff, B., Slovic, P., & Lichtenstein, S. (1978). Fault trees: Sensitivity of estimated failure probabilities to problem representation. Journal of Experimental Psychology: Human Perception and Performance, 4(2), 330–344. https://doi.org/10.1037/0096-1523.4.2.330
Fox, C. R., & Birke, R. (2002). Forecasting trial outcomes: Lawyers assign higher probability to possibilities that are described in greater detail. Law and Human Behavior, 26(2), 159–173. https://doi.org/10.1023/A:1014687809032
Gasper, H. L., Roy, M. M., & Flowe, H. D. (2019). Improving time estimation in witness memory. Frontiers in Psychology, 10, Article 1452. https://doi.org/10.3389/fpsyg.2019.01452
Halevy, N., Maoz, I., Vani, P., & Reit, E. S. (2022). Where the blame lies: Unpacking groups into their constituent subgroups shifts judgments of blame in intergroup conflict. Psychological Science, 33(1), 76–89. https://doi.org/10.1177/09567976211026982
Joint Service Committee on Military Justice. (2025). Amendments to the Military Rules of Evidence since 1984 (G. E. Maggs, Ed.; 2025 ed.). https://jsc.defense.gov/Portals/99/MRE%20Amendments%202025.pdf
Mulford, M., & Dawes, R. M. (1999). Subadditivity in memory for personal events. Psychological Science, 10(1), 47–51. https://doi.org/10.1111/1467-9280.00105
Redelmeier, D. A., Koehler, D. J., Liberman, V., & Tversky, A. (1995). Probability judgment in medicine: Discounting unspecified possibilities. Medical Decision Making, 15(3), 227–230. https://doi.org/10.1177/0272989X9501500305
Rottenstreich, Y., & Tversky, A. (1997). Unpacking, repacking, and anchoring: Advances in support theory. Psychological Review, 104(2), 406–415. https://doi.org/10.1037/0033-295X.104.2.406
Sloman, S., Rottenstreich, Y., Wisniewski, E., Hadjichristidis, C., & Fox, C. R. (2004). Typical versus atypical unpacking and superadditive probability judgment. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30(3), 573–582. https://doi.org/10.1037/0278-7393.30.3.573
Tenney, E. R., Cleary, H. M. D., & Spellman, B. A. (2009). Unpacking the doubt in “beyond a reasonable doubt”: Plausible alternative stories increase not guilty verdicts. Basic and Applied Social Psychology, 31(1), 1–8. https://doi.org/10.1080/01973530802659687
Tversky, A., & Koehler, D. J. (1994). Support theory: A nonextensional representation of subjective probability. Psychological Review, 101(4), 547–567. https://doi.org/10.1037/0033-295X.101.4.547
Legal Authorities
Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579 (1993).
In re Winship, 397 U.S. 358 (1970).
Kumho Tire Co. v. Carmichael, 526 U.S. 137 (1999).
State v. Chapple, 135 Ariz. 281, 660 P.2d 1208 (1983).
State v. Lindsey, 149 Ariz. 472, 720 P.2d 73 (1986).
State v. Nordstrom, 200 Ariz. 229, 25 P.3d 717 (2001).
State v. Portillo, 182 Ariz. 592, 898 P.2d 970 (1995).
United States v. Houser, 36 M.J. 392 (C.M.A. 1993).
Victor v. Nebraska, 511 U.S. 1 (1994).
U.S. Const. amend. XIV, § 1.
Fed. R. Evid. 702.
Mil. R. Evid. 702.
Mil. R. Evid. 1102.
Ariz. R. Evid. 702.
Ariz. Sup. Ct. Order R-23-0004 (amending Ariz. R. Evid. 702, effective Jan. 1, 2024).
Revised Arizona Jury Instructions (Criminal), Standard Criminal 4(a) (Burden of Proof) (State Bar of Arizona, 2022).
Related topic papers: Anchoring in Witness Estimation · The Telescoping Effect · Hindsight Bias · Confirmation Bias · Memory Expert Testimony
Related topics from Arizona Forensic Services: Courts-Martial