Skip to main content

Demoralizing causation

Abstract

There have recently been a number of strong claims that normative considerations, broadly construed, influence many philosophically important folk concepts and perhaps are even a constitutive component of various cognitive processes. Many such claims have been made about the influence of such factors on our folk notion of causation. In this paper, we argue that the strong claims found in the recent literature on causal cognition are overstated, as they are based on one narrow type of data about a particular type of causal cognition; the extant data do not warrant any wide-ranging conclusions about the pervasiveness of normative considerations in causal cognition. Of course, almost all empirical investigations involve some manner of ampliative inference, and so we provide novel empirical results demonstrating that there are types of causal cognition that do not seem to be influenced by moral considerations.

This is a preview of subscription content, access via your institution.

Notes

  1. 1.

    “Normative” here refers broadly to many different types of norms, most notably moral or statistical norms, but also social norms. For ease of exposition, we will throughout subsume these various types of norms under the heading of “normative considerations.”

  2. 2.

    Roughly, some considerations are constitutive of a particular psychological process if and only if they are required inputs for this process, and if as a consequence they are always involved in it.

  3. 3.

    We will focus on everyday causal cognition in this article. It is unclear (even doubtful, in our opinion) whether proponents of the focal thesis intend it to apply to various types of causal inference and reasoning in the sciences, and so we leave that issue aside.

  4. 4.

    This is not a clean division since one can learn new causal information by reasoning about existing knowledge (e.g., the so-called “self-explanation effect”; Chi et al. 1994; Van Lehn et al. 1992).

  5. 5.

    For instance, McGrath (2005, p. 125) notes that “causation is commonly held to be a paradigmatic example of a natural and so entirely non-normative relation.”

  6. 6.

    Normative evaluations could possibly have an impact on low-level processing by changing the individual’s affective state, which then might influence causal perception. Causal perception is incredibly rapid, however, and it seems unlikely that the other processing could occur fast enough to matter. Moreover, even if this theoretically possible “back door path” existed, it would be restricted to normative evaluations that have an emotional impact (e.g., not most evaluations of statistical typicality).

  7. 7.

    We used standard case sequences that have been used in other causal inference experiments. For each sequence, cases were presented in the same pseudo-random order for each participant. Freq(C) = .5 in all sequences, and the conditional frequencies of E were: Generative: Fr(E | C) = .75 and Fr(E | ¬C) = .25; Preventive: Fr(E | C) = .25 and Fr(E | ¬C) = .75; and Non-causal: Fr(E | C) = Fr(E | ¬C) = .5.

  8. 8.

    In order to be able to sensibly present all three sequence-types in the moralized scenario, we used a cover story in which the morally bad actor did not actually know the causal efficacy (or even direction) of his actions, but was so desperate that he tried it anyway.

  9. 9.

    We performed targeted ANOVAs that included only sequence-type, condition, and their interaction; for both experiments, sequence-type was the only significant predictor (Experiment #1: F = 49.3689, p < 10−15; Experiment #2: F = 26.5692, p < 10−9). Tukey HSD post hoc tests showed that all three sequence-types were significantly different from one another, which is relatively clear simply from looking at the mean ratings (Exp. #1: Gen = 45.2, Non-causal = 7.5, Prev = −26.2; Exp. #2: Gen = 49.55, Non-causal = 4.35, Prev = −19.84).

  10. 10.

    The mean blame ratings were 7.87 (Exp. #1) and 5.88 (Exp. #2) on a 1–9 scale, where higher numbers indicate more blameworthy. One complication emerged for Experiment #2: blame ratings differed between the generative (mean = 4.56) and preventive (mean = 7.06) conditions (Tukey HSD post hoc test yields p < .01). This suggests that perhaps both outcome and intentions can matter for blame.

  11. 11.

    We performed ANOVAs for the moralized condition with sequence-type, blame judgment, knowledge judgment, and all interactions. The only significant predictor for either experiment was sequence-type (Exp. #1: F = 14.6268, p < 10−5; Exp. #2: F = 6.3084, p < .005). Note that there was substantial variation in the knowledge ratings (Exp. #1: σ = 2.38; Exp. #2: σ = 2.10).

  12. 12.

    Power calculations are provided in the next section.

  13. 13.

    In ANOVAs for rating given sequence-type and condition, sequence-type was the only significant predictor (Exp. #1: F = 42.6323, p < 10−7; Exp. #2: F = 32.5542, p < 10−6).

  14. 14.

    In an ANOVA for Rating given Scenario and Outcome, there were main effects of both Scenario (F = 26.615, p = .000) and Outcome (F = 43.252, p = .000). A significant main effect of Scenario was also found in each individual Outcome condition (see “Appendix 2”).

  15. 15.

    The 48-case sequences used were:

    Generative: 010203112010310110023110310010210011302110021130

    Neutral: 013220313200213103221013201033121302032103212130

    Preventive: 232021330232132332201332132232032233120332203312 where ‘0’ denotes a C&E case, ‘1’ is ¬C&¬E, ‘2’ is C&¬E, and ‘3’ is ¬C&E.

References

  1. Ahn, W.-K., & Bailenson, J. (1996). Causal attribution as a search for underlying mechanisms: An explanation of the conjunction fallacy and the discounting principle. Cognitive Psychology, 31, 82–123.

    Article  Google Scholar 

  2. Alexander, J., Mallon, R., & Weinberg, J. M. (2010). Accentuate the negative. Review of Philosophy and Psychology, 1, 297–314.

    Article  Google Scholar 

  3. Alicke, M. (1992). Culpable causation. Journal of Personality and Social Psychology, 36, 368–378.

    Article  Google Scholar 

  4. Alicke, M. (2000). Culpable control and the psychology of blame. Psychological Bulletin, 126, 556–574.

    Article  Google Scholar 

  5. Alicke, M., Rose, D., & Bloom, D. (2011). Causation, norm violation and culpable control. Journal of Philosophy., 108, 670–696.

    Google Scholar 

  6. Barbey, A. K., & Sloman, S. A. (2007). Base-rate respect: From ecological rationality to dual processes. Behavioral and Brain Sciences, 30, 241–297.

    Google Scholar 

  7. Blakemore, S.-J., Fonlupt, P., Pachot-Clouard, M., Darmon, C., Boyer, P., Meltzoff, A. N., et al. (2001). How the brain perceives causality: An event-related fMRI study. NeuroReport, 12, 3741–3746.

    Article  Google Scholar 

  8. Buehner, M. J., & May, J. (2002). Knowledge mediates the timeframe of covariation assessment in human causal induction. Thinking & Reasoning, 8, 269–295.

    Article  Google Scholar 

  9. Cheng, P. W. (1997). From covariation to causation: A causal power theory. Psychological Review, 104, 367–405.

    Article  Google Scholar 

  10. Cheng, P. W., & Novick, L. R. (1992). Covariation in natural causal induction. Psychological Review, 99, 365–382.

    Article  Google Scholar 

  11. Chi, M. T. H., De Leeuw, N., Chiu, M.-H., & Lavancher, C. (1994). Eliciting self-explanations improves understanding. Cognitive Science, 18, 439–477.

    Google Scholar 

  12. Cummins, D. D. (1996). Evidence of deontic reasoning in 3-and 4-year-old children. Memory & Cognition, 24(6), 823–829.

    Google Scholar 

  13. Cushman, F., Young, L., & Hauser, M. (2006). The role of conscious reasoning and intuition in moral judgment: Testing three principles of harm. Psychological Science, 17, 1082–1089.

    Article  Google Scholar 

  14. Danks, D. (2009). The psychology of causal perception and reasoning. In H. Beebee, C. Hitchcock, & P. Menzies (Eds.), Oxford handbook of causation (pp. 447–470). Oxford: Oxford University Press.

    Google Scholar 

  15. Danks, D., & Schwartz, S. (2005). Causal learning from biased sequences. In B. G. Bara, L. Barsalou, & M. Bucciarelli (Eds.), Proceedings of the 27th annual meeting of the cognitive science society (pp. 542–547). Mahwah, NJ: Lawrence Erlbaum Associates.

    Google Scholar 

  16. Danks, D., & Schwartz, S. (2006). Effects of causal strength on learning from biased sequences. In R. Sun & N. Miyake (Eds.), Proceedings of the 28th annual meeting of the cognitive science society (pp. 1180–1185). Mahwah, NJ: Lawrence Erlbaum Associates.

    Google Scholar 

  17. Driver, J. (2008a). Attributions of causation and moral responsibility. In W. Sinnott-Armstrong (Ed.), Moral psychology (Vol. 2, pp. 423–440)., The cognitive science of morality: Intuition and diversity Cambridge, MA: MIT Press.

    Google Scholar 

  18. Driver, J. (2008b). Kinds of norms and legal causation: Reply to Knobe and Fraser and Deigh. In W. Sinnott-Armstrong (Ed.), Moral psychology (Vol. 2, pp. 459–461)., The cognitive science of morality: Intuition and diversity Cambridge, MA: MIT Press.

    Google Scholar 

  19. Erev, I., Ert, E., Roth, A. E., Haruvy, E., Herzog, S., Hau, R., et al. (2010). A choice prediction competition, for choices from experience and from description. Journal of Behavioral Decision Making, 32, 15–47.

    Article  Google Scholar 

  20. Evans, J. St. B. T. (2008). Dual-processing accounts of reasoning, judgment, and social cognition. Annual Review of Psychology, 59, 255–278.

    Article  Google Scholar 

  21. Feinberg, J. (1970). Doing and deserving: Essays in the theory of responsibility. Princeton: Princeton University Press.

    Google Scholar 

  22. Fisher, R. A. (1935). The design of experiments. Edinburgh: Oliver and Boyd.

    Google Scholar 

  23. Fugelsang, J. A., Roser, M. E., Corballis, P. M., Gazzaniga, M. S., & Dunbar, K. N. (2005). Brain mechanisms underlying perceptual causality. Cognitive Brain Research, 24, 41–47.

    Article  Google Scholar 

  24. Gigerenzer, G., & Hoffrage, U. (1995). How to improve Bayesian reasoning without instruction: Frequency formats. Psychological Review, 102, 684–704.

    Article  Google Scholar 

  25. Gopnik, A., Glymour, C., Sobel, D. M., Schulz, L. E., Kushnir, T., & Danks, D. (2004). A theory of causal learning in children: Causal maps and Bayes nets. Psychological Review, 111, 3–32.

    Article  Google Scholar 

  26. Haidt, J. (2001). The emotional dog and its rational tail: A social intuitionist approach to moral judgment. Psychological Review, 108, 814–834.

    Article  Google Scholar 

  27. Hart, H. L. A., & Honoré, T. (1985). Causation in the law (2nd ed.). New York: Oxford University Press.

    Book  Google Scholar 

  28. Hau, R., Pleskac, T. J., & Hertwig, R. (2010). Decisions from experience and statistical probabilities: Why they trigger different choices than a priori probabilities. Journal of Behavioral Decision Making, 32, 48–68.

    Article  Google Scholar 

  29. Hau, R., Pleskac, T. J., Kiefer, J., & Hertwig, R. (2008). The description-experience gap in risky choice: The role of sample size and experienced probabilities. Journal of Behavioral Decision Making, 21, 493–518.

    Article  Google Scholar 

  30. Heider, F., & Simmel, M.-A. (1944). An experimental study of apparent behavior. American Journal of Psychology, 57, 243–249.

    Article  Google Scholar 

  31. Hertwig, R., & Erev, I. (2009). The description-experience gap in risky choice. Trends in Cognitive Sciences, 13, 517–523.

    Article  Google Scholar 

  32. Hitchcock, C., & Knobe, J. (2009). Cause and norm. Journal of Philosophy, 106, 587–612.

    Google Scholar 

  33. Knobe, J. (2009). Folk judgments of causation. Studies in the History and Philosophy of Science, 40, 238–242.

    Article  Google Scholar 

  34. Knobe, J. (2010). Person as scientist, person as moralist. Behavioral and Brain Sciences, 33, 315–329.

    Article  Google Scholar 

  35. Knobe, J., & Fraser, B. (2008). Causal judgment and moral judgment: Two experiments. In W. Sinnott-Armstrong (Ed.), Moral psychology (Vol. 2, pp. 441–448)., The cognitive science of morality: Intuition and diversity Cambridge, MA: MIT Press.

    Google Scholar 

  36. Lombrozo, T. (2010). Causal-explanatory pluralism: how intentions, functions, and mechanisms influence causal ascriptions. Cognitive Psychology, 61, 303–332.

    Article  Google Scholar 

  37. Machery, E. (2011). Thought experiments and philosophical knowledge. Metaphilosophy, 4, 191–214.

    Article  Google Scholar 

  38. Machery, E. (2012). Power and negative results. Philosophy of Science, 79, 808–820.

    Google Scholar 

  39. Machery, E. (ms). Evidence and Cognition.

  40. Macpherson, F. (2012) Cognitive penetration of colour experience: Rethinking the issue in light of an indirect mechanism. Philosophy and Phenomenological Research, 84(1), 24–62.

    Google Scholar 

  41. Mandel, D. R. (2003). Judgment dissociation theory: An analysis of differences in causal, counterfactual, and covariational reasoning. Journal of Experimental Psychology: General, 132, 419–434.

    Article  Google Scholar 

  42. McGrath, S. (2005). Causation by omission: A dilemma. Philosophical Studies, 123, 125–148.

    Article  Google Scholar 

  43. Michotte, A. (1946/1963). The perception of causality. London: Methuen.

    Google Scholar 

  44. Pettit, D., & Knobe, J. (2009). The pervasive impact of moral judgment. Mind & Language, 24, 586–604.

    Article  Google Scholar 

  45. Pinillos, N. Á., Smith, N., Nair, G. S., Marchetto, P., & Mun, C. (2011). Philosophy’s new challenge: Experiments on intentional action. Mind & Language, 26, 115–139.

    Article  Google Scholar 

  46. Rakoczy, H., Warneken, F., & Tomasello, M. (2008). The sources of normativity: Young children’s awareness of the normative structure of games. Developmental Psychology, 44(3), 875–881.

    Google Scholar 

  47. Roser, M. E., Fugelsang, J. A., Dunbar, K. N., Corballis, P. M., & Gazzaniga, M. S. (2005). Dissociating processes supporting causal perception and causal inference in the brain. Neuropsychology, 19, 591–602.

    Article  Google Scholar 

  48. Satpute, A. B., Fenker, D. B., Waldmann, M. R., Tabibnia, G., Holyoak, K. J., & Lieberman, M. D. (2005). An fMRI study of causal judgments. European Journal of Neuroscience, 22, 1233–1238.

    Article  Google Scholar 

  49. Schlottmann, A. (1999). Seeing it happen and knowing how it works: How children understand the relation between perceptual causality and underlying mechanism. Developmental Psychology, 35, 303–317.

    Article  Google Scholar 

  50. Schlottmann, A. (2000). Is perception of causality modular? Trends in Cognitive Sciences, 4, 441–442.

    Article  Google Scholar 

  51. Schlottmann, A., & Shanks, D. R. (1992). Evidence for a distinction between judged and perceived causality. Quarterly Journal of Experimental Psychology, 44A, 321–342.

    Article  Google Scholar 

  52. Scholl, B. J., & Tremoulet, P. D. (2000). Perceptual causality and animacy. Trends in Cognitive Sciences, 4, 299–309.

    Article  Google Scholar 

  53. Schulz, L. E., Bonawitz, E. B., & Griffiths, T. L. (2007). Can being scared make your tummy ache? Naive theories, ambiguous evidence and preschoolers’ causal inferences. Developmental Psychology, 43, 1124–1139.

    Article  Google Scholar 

  54. Sloman, S. (2009). Causal models: How people think about the world and its alternatives. Oxford: Oxford University Press.

    Google Scholar 

  55. Stokes, D. (2012). Perceiving and desiring: A new look at the cognitive penetrability of experience. Philosophical Studies, 158(3), 477–492.

    Google Scholar 

  56. Suter, R. S., & Hertwig, R. (2011). Time and moral judgment. Cognition, 119, 454–458.

    Article  Google Scholar 

  57. Sytsma, J., Livengood, J., & Rose, D. (2012). Two types of typicality: Rethinking the role of statistical typicality in ordinary causal attributions. Studies in History and Philosophy of Science Part C, 43, 814–820.

  58. Taylor, R. S., & Chi, M. T. H. (2006). Simulation versus text: Acquisition of implicit and explicit information. Journal of Educational Computing Research, 35, 289–313.

    Article  Google Scholar 

  59. Van Lehn, K., Jones, R. M., & Chi, M. T. H. (1992). A model of the self-explanation effect. Journal of the Learning Sciences, 2, 1–59.

    Article  Google Scholar 

Download references

Author information

Affiliations

Authors

Corresponding author

Correspondence to David Danks.

Appendices

Appendix 1

Experiment 1

Participants

Sixty-three undergraduates at Carnegie Mellon University participated in return for $5. The experiment took approximately 20 min to complete.

Materials and methods

Participants were first provided with a general overview, followed by the Moral and Control scenarios (randomized presentation order). For each scenario, participants were first provided with a cover story, and then observed one of three possible 48-case sequences—either a generative, neutral, or preventive sequence—where the abstract structure was the same regardless of scenario.Footnote 15 To eliminate possible familiarity effects, participants saw different (abstract) sequence-types in the different scenarios, and so there were six different possible conditions (e.g., generative in the Moral condition and preventive in the Neutral condition). The frequency distributions for the three 48-case sequences were:

  • Generative: P(C) = .5; P(E | C) = .75; P(E | ¬C) = .25

  • Neutral: P(C) = .5; P(E | C) = P(E | ¬C) = .5

  • Preventive: P(C) = .5; P(E | C) = .25; P(E | ¬C) = .75

For each sequence, participants observed the cases one at a time, and clicked a button to move to the next case. Each case was described both in text (e.g., “Betrafindalis copernicia: Alive”) and with corresponding images. After observing the full sequence, participants were asked to rate “to what extent [the potential cause] causes [the target effect].” Ratings were collected using a slider that ranged from −100 (“Always prevents”) to +100 (“Always produces”) with an anchor at 0 (“No effect”). The slider moved in increments of 5, so actually corresponded to a 21-point scale. The slider began at 0, but had to be moved before the rating could be submitted (i.e., participants could not simply click through without moving the slider).

The global introduction was:

You are about to be presented with two stories. One story is about a man, Smith, and plants that he is growing in his greenhouse. In the story, a liquid has been applied to the plants but the characters in the story do not know what liquid has been applied to the plants. The liquid may either be a fertilizer, poison or water, and so might lead the plants to flower, die, or it has no effect at all. Your job will be to figure out what liquid was applied to the plants.

You must remember that the relationship between the plant dying or flowering and exposure to the liquid could be quite complicated (if there is any relationship at all!). As an example, there are many plants that are very sensitive to fertilizers, and flower very easily if exposed to them. But, some plants that are very sensitive to fertilizers still might not flower when exposed to them. Likewise, there are many plants that are very resilient and thus do not have serious reactions when exposed to a poison. But plants that are resilient may still have a serious reaction when exposed to a poison.

The other story that you will see is about Johnson, a doctor who has traveled to an island to study the outbreak of a skin disease among a particular group of villagers. Villagers have come into contact with various plants on the island and some have contracted rashes. Your job will be to figure out whether exposure to a certain plant causes the skin disease, makes people healthy, or has no effect at all.

You must remember that the relationship between the rashes (a symptom of the disease) and exposure to the plant could be quite complicated (if there is any relationship at all!). And this skin disease is like many other diseases: different villagers might have different levels of immunity or resistance, and there are likely many different causes of the disease. As an example, there are many people who respond readily to vitamins, and very easily become healthy if they take them. But, in some cases people who normally respond to vitamins may still not become healthy when exposed to them. Likewise, there are many people who are allergic to peanuts, and break out in serious reactions if exposed to them. But, in some cases, people who are allergic to peanuts might not have a serious reaction when exposed to them.

For both stories, you will be presented with a series of individual cases. For each case, you will be shown whether the factor (liquid or plant) was present or absent, and what happened to the flower or person. The factor’s absence will be indicated by a red X over the picture of the factor. These cases will help you figure out whether or not exposure to a particular liquid causes the plant to die, causes it to grow, or has no effect at all; and whether exposure to a particular plant causes villagers to contract a skin disease, causes them to be healthy, or has no effect at all. After viewing all of the pictures, you will be asked to evaluate the causal connection between these factors.

The cover story for the Control scenario was:

Johnson is a doctor traveling to the South Pacific Islands to research the rare skin disease Anthrapora that has been reported on various islands. In particular, she is studying the possible effect of native plants have on the contraction of these diseases. On the island of Tongatapu, Johnson is studying the impact (if any) of Solanaceae delisa on the skin disease Anthrapora. The plant may lead to the skin disease, it may cure the disease, or it may have no real effect at all. It is your job to figure this out.

Johnson interviewed various villagers; some have the local disease, and some do not. She can diagnose villagers as suffering from the skin disease by finding the characteristic rashes. Unfortunately, because of language barriers, the only other information she can get from the villagers is whether or not they have come in contact with the local plant, Solanaceae delisa.

There are thus four different observations Johnson might make: the villager was exposed to the plant and suffers from the disease; the villager was exposed to the plant and is healthy; the villager was not exposed to the plant and suffers from the disease; the villager was not exposed to the plant and is healthy.

You will now see the information – both plant contact and disease status – that Johnson collected for several villagers. After seeing all of the individuals, you will be asked to evaluate the causal connection between the plant and the skin disease on a scale from −100 to +100. Respond with −100 if you think that exposure to the plant (Solanaceae delisa) always prevents the skin disease (Anthrapora). Respond with +100 if you think that exposure to the plant always produces the rashes. And respond 0 if you think the plant is irrelevant for whether the person suffers from the disease or is healthy. Please give your best estimate of the causal strength, even if you are uncertain about what is actually happening.

The cover story for the Moralized scenario was:

Smith is an elderly man who has devoted his life to cancer research. He has been involved in the development of various treatments, which have helped to save the lives of thousands of people. Recently, Smith traveled deep into the Amazon in order to recover a nearly extinct species of orchids called Betrafindalis copernicia.

Betrafindalis copernicia contains a highly concentrated form of the chemical dispofignila. Smith has been experimenting with a synthetic form of dispofignila and has found that it slows down the growth of cancer. While synthetic doses slow the growth down somewhat, only the strongest form of dispofignila, which cannot be synthetically produced and is only found in the orchid Betrafindalis copernicia, slows down the growth substantially, almost to the point of stopping growth altogether. As a matter of fact, Smith was actually able to experiment with one Betrafindalis copernicia. He found that the dispofignila found in the Betrafindalis copernicia was in fact more potent than the synthetic form of dispofignila and that it lead to a significantly greater decrease in cancer growth than the synthetic form of dispofignila.

Betrafindalis copernicias are very rare. They are only found in a remote part of the Amazon, and, because of global warming, only a few dozen plants survive even there. Smith is sure that if he can preserve these plants, then he can develop a cure for cancer. Thus, Smith traveled to the Amazon, returned home safely with all of the orchids, and placed them in his greenhouse.

Smith’s neighbor, Jones, hates Smith. Jones has always despised Smith for no good reason. Jones knows that Smith has recently returned with the only remaining orchids in the world and he wants to kill all of the plants and destroy Smith’s hopes for finding a cure for cancer.

Jones wants to kill the orchids, but he doesn’t want to get caught. He knows that he could simply uproot the plants and kill them, but in order to do that he would have to get into the greenhouse, which is secured by an alarm. However, there is one way that Jones can kill the plants without leaving any evidence. There are several hoses that run a steady flow of water into the greenhouse. Jones knows that if he can inject poison into the hoses, it will kill all of the plants, and no evidence will be left behind.

One night, Jones breaks into an old farmer’s shed and finds several bottles with the label “poison”; Jones grabs one of the bottles and discretely leaves the shed. Unbeknownst to Jones, however, the farmer reuses his bottles: some of the ones labeled “poison” have fertilizers, others simply have water, and of course, some actually do have poison. (The farmer has a system for knowing what each bottle contains, though Jones obviously does not know this system.)

A few days later, Jones fills a syringe with the liquid from the stolen bottle. He then goes to Smith’s greenhouse, pushes the needle through one of the hoses, and injects the liquid into that particular hose.

Your job is to find out whether the liquid that Jones injected into the hose was a poison, a fertilizer, or water (in which case it has no effect). There are thus four different observations you might make: the plant was exposed to the liquid and dies; the plant was exposed to the liquid and produces flowers; the plant was not exposed to the liquid and dies; the plant was not exposed to the liquid and produces flowers.

You will see the relevant information – both liquid contact and whether the plant died or flowered – for several of the plants in Smith’s greenhouse. After seeing all of the plants, you will be asked to evaluate the causal connection between the liquid and the plant flowering on a scale from −100 to +100. Respond with −100 if you think that exposure to the liquid always kills the plant. Respond with +100 if you think that exposure to the liquid always makes the plant flower. And respond 0 if you think the liquid is irrelevant for whether the plant dies or flowers. Please give your best estimate of the causal strength, even if you are uncertain about what is actually happening.

After only the Moralized scenario, participants were asked two further questions with responses on a 9-point scale: a Blame question (“How blameworthy do you think Jones is for attempting to kill Smith’s plants?” from “Not at all blameworthy” to “Extremely blameworthy”) and a Knowledge question (“Do you think that Jones knew what was in the bottle of liquid that he stole?” from “He did not know at all” to “He definitely knew”).

Results and analysis

There were no order effects in the data, so we report results pooling different orders together. The mean ratings are shown in Fig. 1. ANOVA revealed that there was a main effect of sequence-type (F = 49.3689; p = 2.267 × 10−16), but no effect of situation (F = .7742; p = .3807) and no interaction effect (F = 1.9480; p = .147). Tukey HSD post hoc tests showed that all three sequence-types are significantly different from one another (G vs. P: p = .0; G vs. N: p = 2.1 × 10−6; N vs. P: p = 2.34 × 10−5).

Fig. 1
figure1

Mean strength ratings in each condition

Participant ratings in the moralized scenario are perhaps influenced by the participant’s views about blame or knowledge. An ANOVA for Rating with inputs of sequence-type, blame and knowledge judgments, and all interactions showed that sequence-type was the only significant predictor (F = 14.6268; p = 9.532 × 10−6); all other p-values were at least .05. Tukey HSD post hoc tests showed that the Generative sequence-type led to significantly different ratings than the Neutral and Preventive sequence-types in the restricted domain of only the Moral condition, and the latter two sequence-types were close-to-significantly different (G vs. P: p = .0000069; G vs. N: p = .014; N vs. P: p = .0743).

An ANOVA of Blame judgments using sequence-type, knowledge judgments, strength ratings, and all interactions found no significant predictors (all p-values > .12). An ANOVA of Knowledge judgments using sequence-type, blame judgments, strength ratings, and all interactions found a significant effect of only the Sequence-type × Rating interaction (F = 4.5785; p = .01483); all other factors were not significant (all p-values > .10). Specifically, knowledge judgments were negatively correlated with rating for preventive sequences, positively correlated for neutral sequences, and essentially uncorrelated for generative sequences. We have no clear explanation for this small effect.

Experiment 2

Participants

Forty-eight undergraduates at Carnegie Mellon University participated in return for $5. The experiment took approximately 20 min to complete.

Materials and methods

The basic method of Experiment 2 was identical to Experiment 1, and differed only in the cover stories that were used. The global introduction was:

You are about to be presented with two stories. One story is about a man, Jones, and plants that his competitor is growing. In the story, Jones prunes some of the plants, which may help them survive, may kill them, or may have no effect at all.

You must remember that the relationship between the plant living or dying, and the plant being pruned could be quite complicated (if there is any relationship at all!). As an example, there are many plants that require pruning to survive and thrive, while other plants are very sensitive and so die when pruned. And there are many plants that are very resilient and thus do not normally have any significant reaction when pruned (but might in any particular case).

The other story that you will see is about Johnson, a doctor who has traveled to an island to study the outbreak of a skin disease among a particular group of villagers. Villagers have come into contact with various plants on the island and some have contracted rashes. Your job will be to figure out whether exposure to a certain plant causes the skin disease, makes people healthy, or has no effect at all.

You must remember that the relationship between the rashes (a symptom of the disease) and exposure to the plant could be quite complicated (if there is any relationship at all!). And this skin disease is like many other diseases: different villagers might have different levels of immunity or resistance, and there are likely many different causes of the disease. As an example, there are many people who respond readily to vitamins, and very easily become healthy if they take them. But, in some cases people who normally respond to vitamins may still not become healthy when exposed to them. Likewise, there are many people who are allergic to peanuts, and break out in serious reactions if exposed to them. But, in some cases, people who are allergic to peanuts might not have a serious reaction when exposed to them.

For both stories, you will be presented with a series of individual cases. For each case, you will be shown whether the factor (pruning or plant) was present or absent, and what happened to the plant or person. The factor’s absence will be indicated by a red X over the picture of the factor. These cases will help you figure out whether or not pruning causes the plant to die, causes it to grow, or has no effect at all; and whether exposure to a particular plant causes villagers to contract a skin disease, causes them to be healthy, or has no effect at all. After viewing all of the pictures, you will be asked to evaluate the causal connection between these factors.

The cover story for the Control scenario was:

Johnson is a doctor traveling to the South Pacific Islands to research the rare skin disease Anthrapora that has been reported on various islands. In particular, she is studying the possible effect of native plants have on the contraction of these diseases. On the island of Tongatapu, Johnson is studying the impact (if any) of Solanaceae delisa on the skin disease Anthrapora. The plant may lead to the skin disease, it may cure the disease, or it may have no real effect at all. It is your job to figure this out.

Johnson interviewed various villagers; some have the local disease, and some do not. She can diagnose villagers as suffering from the skin disease by finding the characteristic rashes. Unfortunately, because of language barriers, the only other information she can get from the villagers is whether or not they have come in contact with the local plant, Solanaceae delisa.

There are thus four different observations Johnson might make: the villager was exposed to the plant and suffers from the disease; the villager was exposed to the plant and is healthy; the villager was not exposed to the plant and suffers from the disease; the villager was not exposed to the plant and is healthy.

You will now see the information – both plant contact and disease status – that Johnson collected for several villagers. After seeing all of the individuals, you will be asked to evaluate the causal connection between the plant and the skin disease on a scale from −100 to +100. Respond with −100 if you think that exposure to the plant (Solanaceae delisa) always prevents the skin disease (Anthrapora). Respond with +100 if you think that exposure to the plant always produces the rashes. And respond 0 if you think the plant is irrelevant for whether the person suffers from the disease or is healthy. Please give your best estimate of the causal strength, even if you are uncertain about what is actually happening.

The cover story for the Moralized scenario was:

Miracle Works is a cancer research company that investigates the potential effects that various chemicals produced by exotic plants have on cancer cell growth. Recently, a chemical was isolated in a plant called Bertandis Capernicalia. This chemical has been extensively studied and has been shown to reverse the growth of cancer cells. However, the chemical is only produced after the plant has been fully mature for 2 weeks.

The plant was discovered by researchers after they crossed two different plants, Detra Nicalia and Berta Capernica. Researchers soon found out, however, that Bertandis Capernicalia was very hard to keep alive after it was fully matured. Researchers think that pruning might matter, but they do not know exactly how, since the two plants crossed to produce Bertandis Capernicalia—Detra Nicalia and Berta Capernica—respond differently to pruning. Detra Nicalia lives longer if darkish green leaf tips are trimmed off; Berta Capernica dies faster when it is pruned. Thus, pruning might help Bertandis Capernicalia, harm it, or turn out to be just irrelevant.

Jones is a head executive for CFF Treatment Inc., which is a large chemotherapy company. CFF Treatment Inc. makes billions of dollars a year through manufacturing various machines and chemicals that are used in chemotherapy. Jones and the other executives of CFF Treatment Inc. know that if Miracle Works can successfully grow Bertandis Capernicalia, then cancer patients may be able to be effectively treated and cured without ever having to go through chemotherapy.

Jones decides that something must be done in order to prevent Miracle Works from successfully growing Bertandis Capernicalia. Jones decides that Miracle Works must be sabotaged and knows that he must be clever so that he is not caught. He decides that the best way to sabotage Miracle Works (and not get caught) is to prune all of the plants. This way, it will look like one of the Miracle Works employees is responsible for the death of all the plants.

Importantly, Jones has no idea whether pruning will help or hurt the plants. He thinks that it will effectively kill all of the plants, but this is just a guess on his part. Recall that not even the scientists who work at Miracle Works are sure of the effects of pruning of Bertandis Capernicalia.

One night, Jones breaks into Miracle Works and starts to cut the tips off of all Bertandis Capernicalia leaves. A Miracle Works employee walks in and so he is interrupted and must sneak out before he is caught. Due to the interruption, he was only able to prune some of the plants.

You will now see pictures of plants that were randomly chosen from Miracle Works. Some plants have had the tips cut while others have not. Some plants survived, and some did not. Your job will be to determine whether pruning harms, helps, or is irrelevant to Bertandis Capernicalia. After seeing all of the plants, you will be asked to evaluate the causal connection between pruning and the plant surviving on a scale from −100 to +100. Respond with −100 if you think that pruning always kills the plant. Respond with +100 if you think that pruning always makes the plant survive. And respond 0 if you think that pruning is irrelevant for whether the plant lives or dies. Please give your best estimate of the causal strength, even if you are uncertain about what is actually happening.

Suitably adjusted Blame and Knowledge questions were again asked after the causal strength rating was elicited in the Moral situation.

Results and analysis

There were no order effects in the data, so we report results after pooling together different orders. The mean ratings are shown in Fig. 2. ANOVA revealed that there was a main effect of sequence-type (F = 26.5692; p = 8.548 × 10−10), but no effect of situation (F = .0109; p = .9172) and no interaction effect (F = 1.8921; p = .1567). Tukey HSD post hoc tests showed that all three sequence-types are significantly different from one another (G vs. P: p = .0; G vs. N: p = .0000372; N vs. P: p = .042).

Fig. 2
figure2

Mean strength ratings in each condition

Participant ratings in the moralized situation are perhaps influenced by the participant’s views about blame or knowledge. An ANOVA for Rating with inputs of sequence-type, blame and knowledge judgments, and all interactions showed that sequence-type was the only significant predictor (F = 6.3084; p = .00448); all other p-values were at least .28. Tukey HSD post hoc tests showed that the Generative sequence-type led to significantly different ratings than the Neutral and Preventive sequence-types in the restricted domain of only the Moral condition, though the latter two were not significantly different (G vs. P: p = .0033; G vs. N: p = .0172; N vs. P: p = .818).

An ANOVA of Blame judgments using sequence-type, knowledge judgments, strength ratings, and all interactions found only a main effect of sequence-type (F = 3.6638; p = .036; all other p-values > .09). Tukey HSD post hoc tests showed that this was due to Blame judgments for the Preventive sequences being significantly higher than those judgments for the Generative sequences (p = .0096). An ANOVA of Knowledge judgments using sequence-type, blame judgments, strength ratings, and all interactions found only a main effect of sequence-type (F = 6.6906; p = .0034; all other p-values > .08). Tukey HSD post hoc tests showed that Knowledge judgments for the Preventive sequences were significantly higher than for the Generative (p = .00067) and Neutral (p = .0066) sequences, though the latter two were not significantly different (p = .72).

Appendix 2

Participants

A total of 191 participants were recruited through Amazon’s Mechanical Turk. The task took approximately 5 min to complete. Each participant was paid $.15 for participation.

Materials and methods

All participants were randomly assigned to one of the six Scenario-type X Outcome conditions. The global introduction for the Control scenario was:

Johnson is a doctor traveling to the South Pacific Islands to research the rare skin disease Anthrapora that has been reported on various islands. In particular, Johnson is studying on the island of Tongatapu the impact (if any) of a native plant, Solanaceae delisa, on the skin disease Anthrapora. The plant may lead to the skin disease, it may cure the disease, or it may have no real effect at all. It is your job to figure this out.

Johnson interviewed various villagers; some have the local disease, and some do not. She can diagnose villagers as suffering from the skin disease by finding the characteristic rashes. Unfortunately, because of language barriers, the only other information she can get from the villagers is whether or not they have come in contact with the native plant, Solanaceae delisa.

After reading the global introduction, participants were presented with one of the following outcomes:

  1. (1)

    After observing 100 individuals, Johnson notes that most of the people who came into contact Solanaceae delisa had the skin disease Anthrapora, while a few of the people who did not come into contact with Solanaceae delisa had the skin disease Anthrapora.

  2. (2)

    After observing 100 individuals, Johnson notes that a few of the people who came into contact Solanaceae delisa had the skin disease Anthrapora, while most of the people who did not come into contact with Solanaceae delisa had the skin disease Anthrapora.

  3. (3)

    After observing 100 individuals, Johnson notes that a few of the people who came into contact Solanaceae delisa had the skin disease Anthrapora, while most of the people who did not come into contact with Solanaceae delisa had the skin disease Anthrapora.

Participants were then given the following instructions:

You will now be asked to evaluate the causal connection between the plant and the skin disease on a scale from −10 to +10. Respond with −10 if you think that exposure to the plant (Solanaceae delisa) always prevents the skin disease (Anthrapora). Respond with +10 if you think that exposure to the plant always produces the rashes. And respond 0 if you think the plant is irrelevant for whether the person suffers from the disease or is healthy. Please give your best estimate of the causal strength, even if you are uncertain about what is actually happening.

In each case, participants were asked “To what extent does coming into contact with the plant, Solanaceae delisa, cause the skin disease, Anthrapora?” Ratings were made on a scale ranging from −10 to 10.

The global introduction for the Moralized scenario was a follows:

Smith is an elderly man who has devoted his life to cancer research. He has been involved in the development of various treatments, which have helped to save the lives of thousands of people. Recently, Smith traveled deep into the Amazon in order to recover a nearly extinct species of orchids called Betrafindalis copernicia.

Betrafindalis copernicia contains a highly concentrated form of the chemical dispofignila. Smith has been experimenting with a synthetic form of dispofignila and has found that it slows down the growth of cancer. While synthetic doses slow the growth down somewhat, only the strongest form of dispofignila, which cannot be synthetically produced and is only found in the orchid Betrafindalis copernicia, slows down the growth substantially, almost to the point of stopping growth altogether. As a matter of fact, Smith was actually able to experiment with one Betrafindalis copernicia. He found that the dispofignila found in the Betrafindalis copernicia was in fact more potent than the synthetic form of dispofignila and that it lead to a significantly greater decrease in cancer growth than the synthetic form of dispofignila.

Betrafindalis copernicias are very rare. They are only found in a remote part of the Amazon, and, because of global warming, only a few dozen plants survive even there. Smith is sure that if he can preserve these plants, then he can develop a cure for cancer. Thus, Smith traveled to the Amazon, returned home safely with all of the orchids, and placed them in his greenhouse.

Smith’s neighbor, Jones, hates Smith. Jones has always despised Smith for no good reason. Jones knows that Smith has recently returned with the only remaining orchids in the world and he wants to kill all of the plants and destroy Smith’s hopes for finding a cure for cancer.

Jones wants to kill the orchids, but he doesn’t want to get caught. He knows that he could simply uproot the plants and kill them, but in order to do that he would have to get into the greenhouse, which is secured by an alarm. However, there is one way that Jones can kill the plants without leaving any evidence. There are several hoses that run a steady flow of water into the greenhouse. Jones knows that if he can inject poison into the hoses, it will kill all of the plants, and no evidence will be left behind.

One night, Jones breaks into an old farmer’s shed and finds several bottles with the label “poison”; Jones grabs one of the bottles and discretely leaves the shed. Unbeknownst to Jones, however, the farmer reuses his bottles: some of the ones labeled “poison” have fertilizers, others simply have water, and of course, some actually do have poison. (The farmer has a system for knowing what each bottle contains, though Jones obviously does not know this system.)

A few days later, Jones fills a syringe with the liquid from the stolen bottle. He then goes to Smith’s greenhouse, pushes the needle through one of the hoses, and injects the liquid into that particular hose.

After reading the global introduction, participants were presented with one of the following outcomes:

  1. (1)

    Smith enters the greenhouse the next day and does a check-up on the plants. After observing 100 plants, Smith notes that most of the plants that were watered by hose A (the hose that, unbeknownst to Smith, Jones injected the liquid into) were dead, while a few of the plants that were not watered by hose A were dead.

  2. (2)

    Smith enters the greenhouse the next day and does a check-up on the plants. After observing 100 plants, Smith notes that a few of the plants that were watered by hose A (the hose that, unbeknownst to Smith, Jones injected the liquid into) were dead, while most of the plants that were not watered by hose A were dead.

  3. (3)

    Smith enters the greenhouse the next day and does a check-up on the plants. After observing 100 plants, Smith notes that about half of the plants that were watered by hose A (the hose that, unbeknownst to Smith, Jones injected the liquid into) were dead, while about half of the plants that were not watered by hose A were dead.

Participants were then given the following instructions:

You will now be asked to evaluate the causal connection between the liquid and the plant dying on a scale from −10 to +10. Respond with −10 if you think that exposure to the liquid always kills the plant. Respond with +10 if you think that exposure to the liquid always prevents the plant from dying. And respond 0 if you think the liquid is irrelevant for whether the plant dies or not. Please give your best estimate of the causal strength, even if you are uncertain about what is actually happening.

In each case, participants were asked “To what extent does coming into contact with the liquid cause the plant to die?” Ratings were made on a scale ranging from −10 to 10. Additionally people were asked “How blameworthy do you think Jones is for attempting to kill Smith’s plants?” (9-pt scale anchored at 1 = “not at all blameworthy”, 9 = “extremely blameworthy”) and “Do you think that Jones knew what was in the bottle of liquid that he stole?” (9-pt scale anchored at 1 = “he did not know at all”, 9 = “he definitely knew”).

Results and analysis

We began by conducting an ANOVA with Condition and Outcome as predictors of causal ratings. We found a main effect of Condition (F = 26.615, p = .000) and Outcome (F = 43.252, p = .000) and no interaction effect (F = 1.853, p = .160). Importantly, we examined the effects of Condition on each Outcome. For the generative outcome i.e., “most”, we found a significant effect of Condition (F = 5.101, p = .027) with people in the Moralized scenario (M = 6.15, SD = 2.62) making significantly higher causal ratings than those in the Neutral scenario (M = 4.31, SD = 3.79). For the preventative outcome i.e., “few”, we, again, found a significant effect of Condition (F = 5.102, p = .027) with people in the Moralized scenario (M = .500, SD = 5.21) making significantly higher causal ratings than those in the Neutral scenario (M = −1.90, SD = 3.11). Finally, for the irrelevant outcome i.e., “half”, we found a significant effect of Condition (F = 19.544, p = .000) with people in the Moralized scenario (M = 2.60, SD = 3.62) making much higher causal ratings than those in the Neutral scenario (M = −1.77, SD = 4.08).

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Danks, D., Rose, D. & Machery, E. Demoralizing causation. Philos Stud 171, 251–277 (2014). https://doi.org/10.1007/s11098-013-0266-8

Download citation

Keywords

  • Causal judgment
  • Normative considerations
  • Causal reasoning
  • Causation
  • Moral judgment