While religious beliefs are typically studied using questionnaires, there are no standardized tools available for cognitive psychology and neuroscience studies of religious cognition. Here we present the first such tool—the
The online version contains supplementary material available at
In psychological studies of religion, religious beliefs and attitudes are typically assessed by standardized questionnaires and scales, such as the Christian Orthodoxy Scale
Neurocognitive studies typically rely on the accurate measurement of time between the presentation of independent factors, e.g. visual stimuli or auditory tones, and dependent factors, e.g. subsequent behavioural and neurophysiological responses, such as reaction times (RT) or latencies of brain responses such as event related potentials (ERP). In the context of cognitive studies of religion, verbal stimuli coinciding with an individual’s belief system are often expected to be more immediately accessible for cognitive evaluation, yielding faster RT. In this context, RT seems to reflect both the semantic complexity of the religious stimulus and the cognitive processes underlying religious beliefs, and has already been successfully employed in behavioural studies of religious cognition (Cavrak & Kleider-Offutt,
Going beyond cognitive assessment-based research on religion, an increasing number of studies tap into the neural processes underlying religiosity (Brown & Strawn,
Several studies have investigated whether paranormal and religious beliefs also modulate N400 responses (Fondevila et al.,
Currently, there is no validated inventory for carrying out behavioural and/or neuroimaging experiments of Christian beliefs. Such an inventory would allow researchers to measure response times and neurophysiological responses to belief statements and would enable a more mechanistic approach to religious cognition, thereby facilitating interaction between the cognitive science of religion and other fields of cognitive psychology and neuroscience. We propose that such an inventory should meet the following criteria:
Aiming to facilitate behavioural and neuroimaging research of religious cognition, we have developed a new inventory of Christian beliefs, the
In Study 1, we created and tested an extended version of the inventory (480 items), aiming to remove problematic statements before carrying out a confirmatory Study 2 with a shorter final version of the inventory (400 items).
The first version of the inventory consisted of 480 statements, of which 120 related to Christian belief (or lack thereof; to be returned to in more detail below), 120 to moral beliefs (these are generally agreed-upon aspects of good/bad behaviour rather than debated topics such as euthanasia or abortion), 120 to abstract scientific knowledge (specifically scientific statements that are accepted as fact but which a non-specialist individual cannot empirically test, instead relying on expert opinion), and 120 to everyday knowledge (i.e. information that most people acquire through direct experience). These categories were selected to reflect beliefs and knowledge on a scale from individual (personal beliefs) to factual (accepted by all). Religious beliefs are very personal and vary from individual to individual even within a seemingly-singular faith community, moral beliefs are generally agreed upon at a societal level, abstract scientific knowledge is generally accepted as a “fact” but can change as new discoveries are made, and everyday knowledge relates to facts learned from personal experience. The statements were designed to elicit a response detailing the participant’s level of agreement with the statement on a 5-point scale (strongly disagree to strongly agree). Items in the Religious category were created such that each statement is expected to elicit opposite responses from orthodox Christians and Atheists, while items in the other categories are expected to elicit the same responses from both groups.
Statements were created in sets of four, consisting of two “sister” pairs. Within each pair, one version is expected to generate an “agree” response, and one a “disagree” response and between sister pairs the critical-word-to-expected-response pairing is reversed as shown in Tables S1 and 1.
Within each pair, the two sentences were identical except for the critical word, which determined the expected response. All four statements shared a “baseline” segment (the word or words before the critical word), and sister pairs featured the same critical words but framed so that each critical word generated the opposite expected response. This was done to avoid the possibility that responses would be affected by the features of the particular critical word (e.g. word familiarity) through ensuring that all features of critical words (frequency, length, familiarity, etc.) were matched between the Agree and Disagree items. Items in each condition covered a range of topics, as shown in
Critical words were a maximum of three syllables long, and the mean length (in syllables) and frequency (as measured by the SUBTLEX frequency database based on subtitles of British television programs; Van Heuven et al.,
During the development of the CPICB inventory, draft versions were sent to three groups of experts for feedback and comments.
The Religious items list was first sent to lay members of different Christian denominations (Anglican, Pentecostal, Adventist), asking them to report whether they agreed with the individual items provided. Following this, the Religious items list was sent to both Christian and Atheist “experts” who provided feedback on the content of the statements, particularly whether they felt that most Christians/Atheists they encountered day to day would be likely to give a response in the expected direction. These experts included local priests, ministers and priests-in-training from a variety of Christian denominations, including the Anglican, Roman Catholic and Baptist churches in Cambridge, as well as leaders of both national and local atheist/sceptic organisations, such as Humanists UK. The Scientific items list was sent to University of Cambridge scientists working in each of the fields covered (astronomy, physics, Earth science, biology, medicine), who were asked to check the factual content of statements and give their comments. The full list of items (from all conditions) was sent to four native English speakers (a mixture of Christians and Atheists) who provided feedback on the general intelligibility of the items.
Several rounds of revisions were made to the items following feedback from these three groups.
Items were recorded for auditory presentation by a female native English speaker using specialist recording equipment in a soundproof booth. Following recording, sound files were spliced (using Audacity software) so that the main part of the sentence was identical for each pair and the baseline segment (minimum length 110 ms) was identical across all four versions, providing a sufficiently long time window for the 100 ms ERP baseline. The two instances of each critical word were also identical. This ensured that items in any pair differed only in the critical word, and that across sister pairs, the baseline + critical word was identical across Agree and Disagree versions, as shown in Table Sentence pairing and splicing of audio files
In some item pairs the length of the baseline as recorded was naturally longer in one version than another due to the differing onsets of critical words. In such cases, the duration of the baseline was manipulated so as to be somewhere between the two natural lengths, and sound files were carefully spliced so as to maintain the natural rhythm and sound of the sentence while maintaining the splicing formula shown in Table
The main aim of Study 1 was to identify which inventory items successfully elicited the responses we expected from the two groups, in order that we could then select the most successful items for the final list used in Study 2. For this, we specifically recruited participants with strong Christian/Atheist views by advertising our study on the online portal of Cambridge Psychology Research and university advertisement boards, local churches and atheist societies, as well as by sending an invitation to an email list of the Faculty of Divinity at the University of Cambridge and through word of mouth. All participants were aged 18–45 and identified themselves as native speakers of English. Volunteers completed an online screening questionnaire, and 20 participants—10 Christians and 10 Atheists—were selected from amongst those who met four criteria for being either a Christian or an Atheist: (1) self-identification, (2) responses on the Christian Orthodoxy Scale (COS) (Fullerton & Hunsberger,
The selected 20 participants (14 female; age range 18 to 45 years,
Participants signed an informed consent form and took part in a behavioural pilot experiment, for which they were compensated at a rate of £10 per hour. The study protocol was approved by the Cambridge Psychology Research Ethics Committee (approval no.: PRE.2018.040). The study was carried out in accordance with the guidelines of the Helsinki declaration for the treatment of human participants.
The experiment took place in one of the behavioural testing rooms of the Department of Psychology at the University of Cambridge. The room was equipped with a laptop (Apple Inc.) connected to a 22-inch LCD screen, a standard QWERTY keyboard, a pair of headphones, a desk and a chair. The experiment was programmed using MATLAB (MathWorks, Inc.) with the Psychophysics Toolbox version 3 (PTB-3) (Brainard,
At the beginning of the session, participants completed the Edinburgh Handedness Inventory (Oldfield, Visual depiction and instructed meaning of response keys.
A variety of interpretations were given for each response key, as some statements required a response that related to personal opinion whereas some related to factual correctness. Participants were specifically asked to respond according to their views in their “heart of hearts”, not limiting themselves to expressing views that they felt were publicly acceptable, or views endorsed by their church, society or family. We emphasised that we were not looking for what might be seen as the “majority view” of any community that they identified with; we instead wanted to know their unique opinion. For example, it was emphasised that agreeing with the statement “People who believe in X are wrong” is not the same as saying that people should not believe in X, or that one would tell a believer of X that they are wrong.
During testing, participants sat at a distance of 70 cm from the computer screen. A single trial consisted of the presentation of a grey fixation dot on the screen and concomitant delivery of the audio statement through headphones, followed by the presentation of an image representing the five possible response keys (as shown in Fig.
Before commencing the main task, participants completed a practice block of 16 items to become familiar with the protocol. These practice items were designed to reflect the diversity of topics and linguistic structures found in the experimental items. The main task was delivered in 10 blocks of 48 items, which were presented in a randomized order, and participants were encouraged to take breaks between blocks.
All 20 participants completed the task.
To identify the best 25 sets of items (out of 30) in each category (religious, moral, scientific, everyday) to be included in the final set, we calculated for each item how many of the 20 pilot participants had given the “expected” response for their group (either strongly or weakly). Item sets were then ranked by the lowest “expected response” score of any individual item in that set, and lowest-scoring sets were discarded. Where there were a number of sets with the same lowest score, between-condition matching on critical word frequency and critical word length in syllables was taken into account. Of the items selected to remain in the inventory, the lowest number of expected responses to any item (out of 20) was 13 for Religious items (
Following selection of the final list of 400 items, independent Results of Religious vs Moral Religious vs Scientific Religious vs Everyday Moral vs Scientific Moral vs Everyday Scientific vs Everyday
However, we noted that items of Everyday knowledge had a significantly shorter length of critical words in milliseconds and a shorter length of sentences in words than items of the other three categories (see Table
In order to ensure that the splicing process had not yielded sentences with acoustically unnatural properties which could affect participants' behavioural and electrophysiological responses, four linguistically naive native English speakers listened to the 400 selected statements in randomised order and indicated if they detected anything noticeable about the recording. 387/400 items were rated by all raters as sounding completely natural. For 12 items (spread across all four conditions) a splice junction was identified by one rater, and for one item a splice junction was detected by two raters. As these were responses from raters who had been specifically instructed to listen for anomalies in the recordings, we were satisfied that the splicing effects would be negligible when testing participants who were focused on the meaning, rather than the sound quality of the sentences. Following each pilot testing session, participants were prompted to give their thoughts on the task they had just completed, and none mentioned noticing anything about the acoustic quality of the spoken stimuli.
To assess the construct validity of the averaged Religiosity score of the CPICB, the scores were correlated with the summed scores of the Christian Orthodoxy Scale (Fullerton & Hunsberger,
Individual responses to the selected 100 Religious items of the CPICB ranged between −2 (“I disagree”) and +2 (“I agree”). Given that half of the religious statements were designed to elicit Christian Agree/Atheist Disagree responses, while the other half of the religious statements were designed to elicit Christian Disagree/Atheist Agree responses, the average score would have been close to 0 for both groups of participants. We thus reversed the positive or negative sign of Christian Disagree/Atheist Agree statements for all participants, after which a mean score close to +2 indicated that the participant usually responded with a strongly Christian response, i.e. selecting “I agree” for Christian Agree items or “I disagree” for Christian Disagree items. Again, a mean score close to −2 showed more Atheist responses, i.e. selecting “I agree” for Atheist Agree items or “I disagree” for Atheist Disagree items. The absolute values of mean scores denoted the strength of views.
Given that our participants were specifically recruited as having extreme scores (< −45 or > +45) on the Christian Orthodoxy Scale, the data were not normally distributed, so the non-parametric correlation was calculated. We observed a significant positive association between the two measures of religiosity (Spearman’s rho = 0.942, Association between religiosity scores of the Christian Orthodoxy Scale and the Cambridge Psycholinguistic Inventory of Christian Beliefs (
To assess the internal consistency of the CPICB, i.e. whether different sentences measure the same concept, we computed Cronbach’s alpha separately for the Religious, Moral, Scientific and Everyday categories of the inventory. Cronbach’s alpha indicates how closely related a set of items are as a group: < 0.70 is regarded as poor or unacceptable, 0.70–0.79 as fair, 0.80–0.89 as good, and ≥ 0.90 as excellent (Cicchetti,
Given that Cronbach’s alpha is sensitive to missing data, we replaced missing values for individual responses with the mean of the whole group for the Moral, Scientific and Everyday categories, whereas, when replacing missing values in the Religious section, means were calculated separately for Christians and Atheists. Across 400 inventory items and 20 participants (8000 data points), only 42 responses (0.53% of the data) were missing (see
Given that half of the sentences were designed to produce “I agree” and another half “I disagree” responses, inconsistent positive and negative correlations would have distorted the assessment of internal consistency. Thus, before calculating Cronbach’s alpha, the positive/negative sign of Disagree item responses was reversed. Internal consistency of the Cambridge Psycholinguistic Inventory of Christian Beliefs (Study 1) Christian Atheist All 10 10 20 20 20 20 20 100 100 100 100 100 100 400 3 25 0 12 10 37 59 0.96 0.755 0.998 0.954 0.94 0.934 0.987
Encouraged by the preliminary findings of Study 1, we aimed to replicate them with a larger sample of participants. Specifically, given that only relatively strong Christians and Atheists were tested in Study 1, we wanted to extend inventory assessment to more moderate Christians and Atheists. Furthermore, in addition to the assessment of construct validity and internal consistency of the inventory, we aimed to assess its test–retest reliability.
Forty participants were selected from a larger database of > 350 participants willing to take part in studies on religiosity, aiming to represent the full continuum of beliefs from strongly Christian to strongly Atheist, i.e. including those with middle-ground beliefs, unlike in Study 1. As before, participants were recruited by advertising our study on the online portal of Cambridge Psychology Research and university advertisement boards, local churches and atheist societies, as well as by sending an invitation to an email list of the Faculty of Divinity at the University of Cambridge, and through word of mouth. Interested individuals completed the same online screening questionnaire as those taking part in Study 1. Two participants could not complete the second session of the study due to medical reasons and were replaced by another two volunteers.
Ten participants were recruited to each of four groups: Strong-Minded Atheists (Christian Orthodoxy Scale scores of −45 to −72), Moderate Atheists (COS scores of 0 to −45), Moderate Christians (COS scores of 0 to 45) and Strong-Minded Christians (COS scores of 45–72). Contrary to Study 1, participation in Christian practices and self-identification as Christian or Atheist was not taken into account as there was considerable variation among the “moderate” groups, with some participants’ positive or negative direction of COS scores not matching their self-identification, and some considering themselves agnostic or spiritual rather than Atheist or Christian. Among Strong-Minded Christians, participants self-identified as “Anglican” (5), “Newfrontiers” (2), “Evangelical Lutheran” (1), “Catholic” (1) and “Quaker/Anglican” (1). Among Moderate Christians, participants reported belonging to “Anglican” (3), “Protestant” (1), “Salvation Army” (1), “Orthodox” (1) and “Catholic” (2) denominations, whereas two participants did not report denominational identity. Among Moderate Atheists, one reported being “Anglican” and one being “Orthodox”.
A
All participants were aged 18–45 and identified themselves as native speakers of English. A one-way ANOVA to test for group differences showed no significant differences in age (
The study protocol was approved by the Cambridge Psychology Research Ethics Committee (Approval No: PRE.2018.040), and it was carried out in accordance with the guidelines of the Helsinki declaration for the treatment of human participants. Participants were compensated at a rate of £10 per hour.
Based on the findings of Study 1, the final version of the inventory used in Study 2 consisted of 400 statements of which 100 related to Christian beliefs (or lack thereof), 100 to moral beliefs, 100 to scientific abstract knowledge and 100 to everyday knowledge, which were selected from the first version of the inventory consisting of 480 statements (see Study 1). The remaining statements were kept in the original sets of four, consisting of two “sister” pairs (see Table
Individual participants took part in two sessions of behavioural testing, which were carried out by the same experimenter using the same equipment in the same lab. We planned the delay between visit 1 and visit 2 to last between two weeks and two months. Most of the participants preferred the shortest possible delay and there were on average 18.5 days (
As for Study 1, the experiment took place in one of the behavioural testing rooms of the Department of Psychology at the University of Cambridge. The room was equipped with a laptop (Apple Inc.) connected to a 22-inch LCD screen, a standard QWERTY keyboard, a pair of headphones, a desk and a chair. The experiment was programmed using MATLAB (MathWorks, Inc.) with the Psychophysics Toolbox version 3 (PTB-3) (Brainard,
At the beginning of the first session, participants signed informed consent forms and completed the Edinburgh Handedness Inventory (Oldfield,
During the second session, participants read the same instructions about the study and were talked through the possible responses to the sentences they would hear in the main task in the same way as they were introduced to the task during the first session. Afterwards, participants carried out the practise block with the same 16 sentences, followed up by the main task delivered in 10 blocks of 40 items, and they were again encouraged to take breaks between blocks. Items were presented in a randomized order, which was generated independently of the first session. Once participants completed the task, the experimenter asked them to fill in the Cognitive Reflection Test (Frederick,
The data analysis plan for Study 2 was preregistered at the OSF Registries, as described in the following. For more details, see
To assess the construct validity of the averaged Religiosity score of the CPICB, it was correlated with the summed scores of the Christian Orthodoxy Scale (Fullerton & Hunsberger,
Given that the 40 participants were specifically recruited to cover different ranges of the Christian Orthodoxy Scale, their Christian Orthodoxy Scale summed scores had a platykurtic distribution with flat tails (excess kurtosis = −1.52) and did not appear to be normally distributed (Shapiro-Wilk test:
To assess the internal consistency of the CPICB, i.e. whether the 100 sentences within each category measure the same concept, we computed Cronbach’s alpha separately for the Religious, Moral, Scientific and Everyday sections of the Inventory and for the full inventory. Cronbach’s alpha < 0.70 was regarded as poor or unacceptable, 0.70–0.79 as fair, 0.80–0.89 as good and ≥ 0.90 as excellent (Cicchetti,
Given that Cronbach’s alpha is sensitive to missing data, we replaced missing values for individual responses with the mean of the whole group (
Given that half of the sentences were designed to produce “I agree” and another half “I disagree” responses, inconsistent positive and negative correlations would have distorted the assessment of internal consistency. Thus, before calculating Cronbach’s alpha, the negative/positive sign of Disagree items was reversed for the Moral, Scientific and Everyday sections of the inventory. Likewise, the negative/positive sign of Christian Disagree/Atheist Agree items was reversed for the Religious section of the inventory. Items with identical responses from all participants were removed from the scale before calculating Cronbach’s alpha.
To assess test–retest reliability of the mean scores within each category of the inventory (Religious, Moral, Scientific and Everyday), a two-way mixed model intraclass correlation coefficient of an absolute agreement type (ICC 3.1) was used to compare performance on sessions 1 and 2 (Koo & Li,
For each session, a mean response score was calculated separately for the Religious, Moral, Scientific and Everyday sections of the inventory. Given that half of the statements were designed to elicit Agree responses, while the other half of the statements were designed to elicit Disagree responses, the average score would be close to 0 for all groups of participants. We thus have reversed the positive or negative sign of (Christian) Disagree statements for all participants. In the Religious category, a mean score close to +2 indicated that the participant usually responded with a strongly Christian response, whereas a mean score close to −2 showed more Atheist responses. In addition to the mean scores over both agreement conditions, we reported test–retest reliability statistics separately for the (Christian) Agree and (Christian) Disagree items.
We observed a significant positive correlation between the COS and the CPICB measures of religiosity (Spearman’s rho = 0.915, Association between religiosity scores of the Christian Orthodoxy Scale and the Cambridge Psycholinguistic Inventory of Christian Beliefs (
Across 400 inventory items and 40 participants (16000 data points), only 82 responses (0.51% of data) were missing (see
We observed excellent internal consistency (Cronbach’s alpha ≥ 0.90) at a full group level ( Internal consistency of the Cambridge Psycholinguistic Inventory of Christian Beliefs (Study 2) Strong- Minded Atheists Moderate Atheists Moderate Christians Strong- Minded Christians All 10 10 10 10 40 40 40 40 40 100 100 100 100 100 100 100 100 400 0 0 0 0 0 2 2 14 18 0.95 0.94 0.98 0.57 1 0.93 0.92 0.90 0.98
For one participant, most of the trials were registered as unresponsive during Session 2 (377 out of 400), most likely due to a keyboard malfunction. Hence, we excluded this participant from the test–retest reliability analysis, with a remaining total of 39 participants. When comparing the mean scores of the CPICB between Session 1 and Session 2, ICC values indicated excellent test–retest reliability for the Religious and Moral categories, good reliability for the Scientific category, and fair reliability for the Everyday category (see Table Test–retest reliability of the Cambridge Psycholinguistic Inventory of Christian Beliefs CPICB category Intraclass correlation 95% Confidence interval Religious .974 .951 .986 74.860 38 2.7E−26 Moral .843 .721 .914 11.696 38 4.7E−12 Scientific .747 .566 .859 6.766 38 2.1E−8 Everyday .524 .252 .719 3.161 38 0.0003 Association between the mean scores of the Cambridge Psycholinguistic Inventory of Christian Beliefs categories obtained on session 1 and session 2 (
Here we present a new instrument—the
The CPICB shows robust psychometric properties, including its construct validity, internal consistency and test–retest reliability. Mean responses to the CPICB religious (dis)belief statements were very strongly associated with the mean scores of the Christian Orthodoxy Scale (Fullerton & Hunsberger,
While there was excellent internal consistency for each category of the CPICB when testing a full group of participants (
A major strength of the CPICB is that statements are carefully controlled and items are closely matched across conditions, with each item being presented in almost identical “agree” and “disagree” versions, and identical critical word lists used across agree and disagree versions (see details in Study 1, Methods). These measures make the CPICB suitable for neurocognitive studies where such matching is necessary in order to isolate the brain response to a specific difference between two sentences while also reducing the variation between conditions that might otherwise occur as a result of a participant having different levels of familiarity with the content or words of particular items. However, future researchers should be aware that while the sets of critical words have been matched across conditions for word frequency, intensity (dB) and length in syllables (see Table
The advantage of psychoacoustic and linguistic matching of long statements comes at the cost that it can be difficult to construct sets of sentences (such as the sister-pair sets shown in Table
Another area where the creation of perfectly matched sentences has had to be balanced with naturalness is in the splicing of the audio files. As files are spliced so that items in each condition are identical up to the critical word, a researcher can be confident that any response differences across conditions are due to the critical word alone, and not to any minor intonation/volume/emphasis differences in the main part of the sentence. However, due to the way that the pronunciation of a word in connected speech is influenced by the words preceding and following it, in some cases it required compromise in order to ensure that the baseline section had the same duration over all four statements in a sister-pair set. For example, due to catenation between a final consonant and an initial vowel in connected speech, the “is” in “is wise” is pronounced differently from the “is” in “is unwise”, as in the latter there is no gap between the two words. In such cases we digitally manipulated the baseline in order to find the best fit that allowed an identical baseline to be used in both versions of the sentence, while retaining the natural rhythm and sound of both. As our pre-experiment checks showed that naive participants were occasionally able to identify a splice junction, this raises the question of whether participants’ natural processing could be influenced by (possibly subconscious) awareness of these minor changes to the natural timing of the sentence. While this may be an issue in participants who have direct experience in splicing audio materials or are otherwise hyper-aware of connected speech patterns, the fact that 387/400 items were rated by all 4 raters as sounding completely natural satisfies us that this is unlikely to be an issue in naive participants. In addition, all participants in both study 1 and 2 were asked for their impressions at the end of the experiment, and none reported detecting anything anomalous with regard to the naturalness of the audio files.
While listening to a sentence, certain expectations naturally build up regarding how the sentence will end. Kutas and Hillyard (
One of the key limitations of the CPICB is its narrow focus on only one dimension of religiosity—explicit religious beliefs—among Christians. Relatively recent developments in the study of Christianity (or, as some scholars are careful to specify, the study of
The sheer diversity of doctrinal beliefs within the various expressions of Christianity worldwide make it difficult, if not impossible, to speak of “Christian belief” as if it were a homogeneous and monolithic phenomenon. The particular theological viewpoints which informed our understanding of Christian beliefs were located within the cultural and theological milieus of Cambridge, UK, which is a progressive university town. There are a range of Christian theological viewpoints which were not explicitly foregrounded in our questions and it might very well be the case that an individual who strongly self-identifies as a Christian would not score as such upon using our inventory. One wonders, for example, if Christians whose theological and cultural contexts are significantly different from those of the Christians who guided our statements (see, for examples of strikingly different Christian theologies, the ethnographic accounts of Bauman (
With the above-mentioned caveats in mind, we nonetheless feel that it has been important and valuable to focus on individuals’ responses to statements of (religious) belief, and to construct those statements of belief with reference to the specific theological milieus that we did. Pragmatically, it will be easier to carry out quantitative group-level studies with participants committed to fundamental theologies, which allows us to predict participants’ answers and plan group comparisons. Contrary to this, participants recruited from communities with more liberal or less defined theologies would likely provide very diverse answers to the same questions, making it very hard to contrast groups of religious and non-religious individuals.
Finally, even though we observed an excellent internal consistency of responses to religious statements at a full group level in both studies, lower consistency within smaller subgroups suggests that the inventory may cover beliefs that yield varying responses within and across individuals. When developing the CPICB, we aimed to cover five broad categories of Christian beliefs, namely anthropological beliefs, God attributes, prophecies and eschatology, supernatural agents, and miracles (see
We conclude that the CPICB is a reliable tool and we hope it will facilitate new research lines in the experimental psychology and cognitive neuroscience of religion. The CPICB is available (open access) on the project’s OSF site:
The study was supported by the John Templeton Foundation (grant ID 60936).
(DOCX 29 kb)
The qualifier ‘Orthodoxy’ within the ‘Christian Orthodoxy Scale’ refers to the
This question requires participants to indicate which of seven statements comes closest to expressing what they believe about God:
1.I don't believe in God.
2. I don't know whether there is a God, and I don't believe there is any
way to find out.
3. I don't believe in a personal God, but I do believe in a Higher Power
of some kind.
4. I find myself believing in God some of the time, but not at others.
5. While I have doubts, I feel that I do believe in God.
6. I know God really exists and I have no doubt about it.
7. Don't know.
In the Project registration, this information is not clearly presented, and it can be understood that positive/negative signs were reversed only for the Christian Disagree/Atheist Agree items in the Religious category. We confirm that positive/negative signs were also reversed for all Disagree items in the Moral, Scientific and Everyday categories.
Results of the Cognitive Reflection Test will be reported elsewhere.
Open Practices Statement
The data and materials for all experiments are available at
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.