Edited by:
Reviewed by:
*Correspondence:
This article was submitted to Cognition, a section of the journal Frontiers in Psychology.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
In determining the prime cause of a physical event, people often weight one of two entities in a symmetric physical relation as more important for bringing about the causal effect than the other. In a broad survey (
“Is the fact that a piece of wood floats on water basically due to(a) the piece of wood or (b) the water?”
Confronted with this question, about 70% of the German participants in two studies respond “it is due to the wood” (
This kind of weighting in symmetric physical settings is not unusual; we find it for different physical events and in different cultures (e.g., ;
This plurality of factors resonates quite well with what we know from research on causal judgments in general, namely that such judgments can be influenced in various ways: by the familiarity, imageability, and believability of cause and effect relationships (
In this paper we focus specifically on one modulating factor:
When thinking about whether the piece of wood or the water is responsible for the floating, people establish a “causal field” (
Different kinds of processes contribute to this foregrounding: in some cases, one entity catches people’s attention due to particularly salient features (e.g.,
Some of these foregrounding processes build on an explicit reference to a contrasting situation (cf.
Our survey on causal judgments in symmetric physical settings (
The findings from
This finding, however, is restricted in three ways. The first concerns the breadth of the results. From the nine physical settings included in our 2011 survey, only one was devoted to this kind of contrast effect. The second restriction is more severe as it concerns how we assessed people’s causal assignments. With one exception, all of our tasks in that survey used the forced-choice format as illustrated in the examples above. The main reason for doing so was that we wanted to prevent participants “from simply activating their formal knowledge acquired in school” (
In the following, we present two experiments that aim at generalizing and validating the contrast effect from our previous survey (
We hypothesize that the contrast effect generalizes to different physical settings. If participants take the contrasting situations into account, this should affect their causal assignments as compared to a baseline task without such a contrast, and it should be reflected in how they explain the physical event in question. To get a spontaneous evaluation of which factor is considered to be the main cause, we first ask for a causal assignment, and then for an explanation of the decision. The response format might have a differentiating effect: participants’ causal assignments might overall be less biased with the rating format than with the forced-choice format as only the former allows them to consider the relational character of the task appropriately. This might also be reflected in people’s explanations.
Finally, in order to be able to assess the extent to which possible content-specific associations with the involved entities (wood/water and CO2/air, respectively) contribute to participants’ assignments and give rise to
This experiment aimed at generalizing the contrast effect from
A total of 111 students (30 male, 81 female) from the University of Freiburg (Germany) volunteered to participate in the experiment. The mean age was
Six causal assignment tasks were constructed by crossing three conditions with different contrast situations (no contrast, contrasting object vs. contrasting liquid) with two content scenarios (concrete vs. abstract). All tasks referred to the same physical relation (an object floating on a liquid) and called for a forced-choice decision on which of the two entities involved is responsible for the floating.
The three concrete tasks read as follows:
□ the piece of wood.
□ the water.
□ the piece of wood.
□ the water.
□ the piece of wood.
□ the water.
The corresponding abstract tasks read as follows:
□ the object.
□ the liquid.
□ object B.
□ the liquid.
□ the object.
□ liquid B.
Participants were instructed to make their assignment spontaneously and to mark one option. Each task was followed on a separate page by the request to briefly explain their decision (“Bitte begründen Sie kurz Ihre Entscheidung.”). All materials were presented in German and were pretested for comprehensibility.
The experiment was part of a larger paper-and-pencil questionnaire study with different kinds of reasoning tasks (none of the other tasks, however, was on a causal topic). The part of the questionnaire relevant here consisted of four pages: (i) first causal assignment task, (ii) first explanation, (iii) second assignment task, and (iv) second explanation.
For each questionnaire booklet, one of the three concrete contrast conditions was paired with one of the three abstract contrast conditions (in first or second position, respectively). From the 18 possible combinations of conditions, we implemented all possibilities except those with a no-contrast condition in the second position. The reason for this restriction was that we wanted to prevent a possible carry-over of contrast information from the more complex conditions with a contrasting situation to the simpler no-contrast conditions. We thus had six pairings with a concrete task in first position and six pairings with an abstract task in first position. The order of the answer options (object first vs. liquid first) was balanced within the pairings (across the two tasks) and also across the pairings. We prepared an appropriate number of the different booklet versions so that, in the end, the different conditions of both the concrete and the abstract scenario occurred with (almost) equal frequencies across participants. Finally, participants were assigned randomly to the prepared questionnaires.
Participants’ causal assignments are presented in
A four-way log-linear analysis was conducted (
The concrete tasks revealed the following results: in the no-contrast condition, participants preferred the wood as causative for the floating (68.4% wood vs. 31.6% water; χ2 = 5.2,
An analogous pattern was found for the abstract tasks: in the no-contrast condition, participants again preferred the object as causative for the floating (76.3% object vs. 23.7% liquid; χ2 = 10.5,
These findings replicate the biased causal assignments for the relation “wood floats on water” in the absence of explicit contrasts (as described by
In order to shed light on participants’ rationale for their decisions, we had asked them for explanations. Their statements were classified into four categories: “relational,” “property,” “contrast,” and “others.” A statement was classified as
The explanation data revealed two main findings. First, in the no-contrast tasks, participants gave all three types of explanations with similar frequencies. Explanations in terms of individual properties of floater and liquid slightly prevailed, mostly referring to density and weight. This also means that some participants used the contrast strategy to explain the floating without being triggered by an explicit contrast situation. Second, compared to this baseline, relational explanations decreased from 32.9 to 19.2% when a contrasting situation was introduced (χ2 = 5.2,
All in all, Experiment 1 was successful in extending the contrast effect on a new symmetric physical setting. However, as in our previous studies, we used a response format that forced participants to take a dichotomous decision. With this format, we intended to outmaneuvre possible effects of school education and to tap into deeper, folk-theoretical convictions, but it could be argued that this format also has skewed participants’ responses in an unintended way and beyond their actual preferences. This concern is addressed in the next experiment.
In order to assess the influence of the response format, Experiment 2 combines the original physical setting from
A total of 128 students (32 male, 96 female) from the University of Freiburg (Germany) volunteered to participate in the experiment (there was no overlap with participants from Experiment 1). The mean age was
Six causal assignment tasks were constructed by crossing three conditions with different contrast situations (no contrast, contrasting floater vs. contrasting medium) with two content conditions (concrete vs. abstract). All tasks referred to the same physical relation (a gas staying down in a substance) and allowed participants to indicate the relative causal effectiveness of the entities involved by means of an analog rating scale of 10 cm length.
The three concrete tasks read as follows:
the CO2 |———————————————| the air.
the CO2 |———————————————| the air.
the CO2 |———————————————| the air.
The corresponding abstract tasks read as follows:
gas G |———————————————| substance S.
gas G |———————————————| substance S.
gas G |———————————————| substance S.
Participants were instructed to make their assignment spontaneously. Each task was followed on a separate page by the request to briefly explain their decision from the previous task (“Bitte begründen Sie kurz Ihre Entscheidung aus der vorigen Aufgabe.”). All materials were presented in German and were pretested for comprehensibility.
These were analogous to Experiment 1. However, instead of the order of the response options, the polarity of the rating scale was balanced (CO2/gas G marking the left side vs. the right side of the scale).
Participants’ causal assignments were coded by measuring their marks on the rating scale (accurate to 0.5 mm) ranging from 0 (the mark was right on the endpoint labeled with “air/substance”) to 10 (the mark was right on the endpoint labeled with “CO2/gas G”). Therefore, values above 5.0 indicate a stronger causal role of the floater (CO2/gas G), and values below 5.0 indicate a stronger causal role of the medium (air/substance S). The results are presented in
In the first step, we checked the rating data for possible position effects. To this end, two analyzes of variance were conducted (one for the concrete tasks and one for the abstract tasks) with “position of task” (first vs. second position) and “contrasting situation” (contrasting floater vs. medium) as independent variables. We did not include the no-contrast conditions in these analyzes as the respective tasks were presented always in first position. As the two analyzes did not indicate any position effects (for all effects with “position”:
In the second step, we analyzed the ratings with “contrasting situation” (now all three types) and “polarity of the rating scale” as independent variables, again separately for concrete and abstract tasks. The two analyzes did not indicate any polarity effects [for all effects with “polarity”:
The concrete tasks exhibited the following results: in the no-contrast condition, we found an imbalanced causal assignment: participants preferred the gas (CO2) over the medium (air), with
A slightly different pattern was found for the abstract tasks: this time, the no-contrast condition indicated a quite symmetric causal assignment with
Next, we categorized people’s explanations of their causal assignments as described for Experiment 1. The results are shown in
The increase of relational explanations, and correspondingly the decrease of the other types of explanations, is most probably a direct consequence of the differences in the response format (as is the overall less extreme causal assignments compared to Experiment 1): with the forced-choice format, participants have to commit themselves to one entity as causative in the assignment task, and later on have to explain this decision regardless of whether or not they are able to figure out the physically appropriate response. With the rating-format, however, participants can indicate in the assignment task that both entities are equally causative, and are therefore not restricted later on to explain an either-or decision, but might draw on their physical knowledge instead.
If we assume that those participants providing an adequate relational explanation did not base their (preceding) causal assignment on a contrasting situation, while the other participants did include a contrast in their reasoning in one or the other way, then we would expect the contrast effect to disappear for the first group and to be more pronounced for the second. To test this prediction, we re-analyzed the ratings by means of two analyzes of variance (one for the concrete tasks and one for the abstract tasks) in which we included the “type of explanation” (relational vs. all other) as an independent variable in addition to the “contrasting situation” and the “polarity of the rating scale.” As before, these analyzes revealed no polarity effects, but main effects of the contrasting situation [concrete content:
For the concrete tasks, we did not find any effect of the type of explanation [largest
For the abstract tasks, the interaction “type of explanation × contrasting situation” was significant [
On the whole, Experiment 2 was successful in validating the contrast effect with a less restrictive response format, and the inspection of people’s explanations provided further evidence for the assumed contrasting mechanisms. Participants’ causal assignments appeared to be fairly well reflected in their later explanations, but the correspondence was not perfect. In about 30% of the cases, in which participants apparently
Physical settings in which two entities interact to bring about an effect are symmetric, and, from that point of view, responsibility should be assigned in a balanced manner. Often, however, people make imbalanced causal assignments by giving one of the factors a stronger weight. Against this background, the goals of this paper were twofold: we wanted to validate the method of assessing such weighting effects, and to put the contrast effect from
Taken together, the experiments yielded four main results: first, in three of the four conditions without an overt contrast we observed biased causal assignments. These were more pronounced in Experiment 1 with the forced-choice format than in Experiment 2 with the rating format, and they replicate previous findings quite well (
Second, the response format clearly made a difference. Giving participants the possibility to indicate the relative causal effectiveness of the two entities in question, instead of forcing them into a decision, resulted generally in more balanced causal assignments and more relational explanations. This does not imply, however, that all biases automatically disappear with this kind of assessment. The concrete no-contrast task in Experiment 2 is a case in point, showing a small but nevertheless stable imbalance (see
Third, irrespective of the response format, the two experiments confirm that people are sensitive to contrast information. Experiment 2 replicates the general data pattern from
Finally, the results also provide some indication for content effects (
Two possible limitations need to be addressed though: first, we had instructed our participants to make their causal assignments
Finally, two broader questions remain to be answered, a normative one and a psychological one: should we consider the contrast effect in symmetric physical settings a
The first question concerns the tension between what (many) people actually
The second question on how the imbalanced causal assignments in the no-contrast conditions can be explained is more difficult to answer. Here, participants’ explanations—albeit as
Taken together, this diversity of explanations suggests that biases in causal assignments are (likely) triggered not by one single rationale, but can be triggered by a host of different cues. The general mechanism involves processes of activating, selecting, and weighting reasons, but we have only just begun to explore this conglomeration—by singling out and varying the influence of some factors or, in other words, by contrasting situations.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This work was supported by the Deutsche Forschungsgemeinschaft DFG through a grant for the project “Die Zuweisung kausaler Rollen in der physikalischen Domäne” to Sieghard Beller and Andrea Bender (Be 2178/5-1). All experiments were carried out at the University of Freiburg (Germany) and conformed to the relevant ethical standards. We are grateful to Miriam Hansen and Nikol Rummel for support with data collection for Experiment 1, Susanne Bubser and the Social Psychology and Methodology group for support with data collection for Experiment 2, and to Susanne Bubser, Sarah Mannion de Hernandez as well as Annelie Rothe-Wulf for valuable comments on earlier versions of this article. We also thank the Center for Interdisciplinary Research (ZiF) at the University of Bielefeld (Germany) for providing an inspiring research environment, and the members of the ZiF research group “The cultural constitution of causal cognition” for numerous stimulating discussions on this topic. The article processing charges were covered by the Publication Fund of the Library of the University of Bergen.
A broader discussion of causal reasoning across domains and cultures is provided in the Research Topic
As the baseline tasks were only presented in first position, four “zero cells” were defined to accommodate the fact that not all possible combinations of the independent variables were instantiated.
A log-linear model fits the data if the test statistic does