1 Reply

Bullshit has been defined as something that is constructed without concern for the truth (Reference FrankfurtFrankfurt, 2005). By this definition, bullshit statements can be true, false, or meaningless. The absence or presence of these factors is irrelevant to something being bullshit. Nonetheless, although bullshit statements can be incidentally true, bullshit is generally false and hence, often problematic.

In our initial investigation of bullshit, we focused on statements that consisted of randomly selected buzzwords (Reference Pennycook, Cheyne, Barr, Koehler and FugelsangPennycook et al., 2015). We used 20 different statements across 4 different studies (excluding items from Deepak Chopra’s Twitter feed, which we will not discuss further in this reply). Examples include “wholeness quiets infinite phenomena” (wisdomofchopra.com) and “we are in the midst of a high-frequency blossoming of interconnectedness that will give us access to the quantum soup itself” (sebpearce.com/bullshit). We labelled these statements “pseudo-profound bullshit” because: 1) they were constructed absent any concern for the truth and, generally, for that reason, 2) they do not consistently have unambiguous meaning, though they can sometimes interpreted by at least some people to have profound meaning.

In his commentary, Dalton notes, as do we, that at least some randomly generated statements can be taken as meaningful by some readers. Dalton takes this claim to have methodological implications but we will argue below this cannot be true without further untenable assumptions. We take Dalton’s claim to actually be conceptual, if not philosophical. Specifically, Dalton’s conceptual point appears to be based on a radical reader-response theoretical position in which the meaning of a text is solely what the reader makes of it. From such a perspective it is, of course, not possible to say, a priori, that any text is ultimately and always meaningless as some reader will always have the last say. Dalton argues, it is (and will always be) possible for at least someone, somewhere, to find (or perhaps more aptly “construct”) meaning, or what they take to be meaning, by sufficiently contemplating any statement. We, of course, agree as the very premise of our study was that people would report sentences designed without regard to meaning to be at least somewhat profound. Bullshit, following the definition offered by Frankfurt, however, depends on the intentions (or lack thereof) of the person uttering or writing the relevant statements. Bullshit that is viewed as profound is still bullshit.

As a consequence, without endorsing a radical reader-response theory, we note that Dalton’s primary point is consistent with the goal of our study. Namely, we hypothesized that people would indeed report randomly generated statements as not only meaningful but profound and, moreover, that people would vary in this propensity. This expectation was based on the assumption that people will find, or suppose that they have found, meaning in such statements. The very goal of the study was to investigate this tendency empirically, not to argue, as Dalton states, that: “if one cannot immediately discern meaning in something it is automatically bullshit.” It is important to recognize that even if we take a radical reader-response position, the only constraint on our original study was in the use of “pseudo-profound” as a label for the random sentences. Because they were constructed without any concern for the truth, they are bullshit (by the Frankfurt definition we followed in our original paper). Dalton appears to assume that our use of the term bullshit implies some sort of value judgment, as it often is in everyday use. In contrast, and following Frankfurt’s lead, we used bullshit as a technical term. This is the way that we hope it continues to be used in the academic literature.

With regard to Dalton’s claim that his argument is methodological, we note that this cannot be so unless one assumes that a sample of random computer-generated statements have an equal probability of being interpreted as meaningfully profound as human-generated statements intended to be meaningful (if not profound). Methodologically, the key word in the forgoing sentence is sample. In our study we found and reported that a sample of human-generated profound quotations (e.g., “A wet person does not fear the rain”) were rated as more profound than samples of computer-generated random sentences (see Studies 3 and 4). Thus, the equality of meaning assumption was demonstrably false for our study.

In reference to the “Wholeness quiets infinite phenomena” example, Dalton takes a phenomenological stance to meaning: “To engage with a passage like this we need to contemplate it for more than a few seconds, perhaps a few minutes (or hours, days, or months) and watch what happens to our mind – this is the appropriate first person subjective experience and more appropriate outcome of interest.” In the research under discussion, however, such an approach is not only inappropriate, but altogether irrelevant. Our interest, in this initial study, was not in the first-person phenomenology of readers’ subjective experience (though it might constitute a possible and interesting subsequent line of research) but simply in participants’ profundity ratings of statements designed to be lacking in that very quality (i.e., as a way to index one’s receptivity to bullshit). Dalton’s argument does imply the interesting and plausible hypothesis that people who are more reflective, clever, and/or linguistically adept will be more apt to construct meaning in the ambiguous statements. Unfortunately for this hypothesis, increasing reflectivity was negatively associated with reporting greater profundity in bullshit statements. One possibility is that more reflective people were indeed more able to find meaning where none was intended but are also likely to realize that the meaning was constructed through their own cognitive efforts rather than by the ostensible author of the statements.

One potential response to this line of reasoning is that it may be that only some of the randomly generated statements are potentially meaningful to some readers or, perhaps more precisely, that the statements varied in the ease with which they could be assigned some meaning. Dalton notes, for example that he cannot derive meaning from the following statement: “We are in the midst of a high-frequency blossoming of interconnectedness that will give us access to the quantum soup itself”, though we suspect that some people might. This possibility suggests the further hypothesis that the association between our variables of interest and profundity ratings for the bullshit items might vary as a function of the ease of constructing profound meaning from randomly generated sentences. To test this, we created two new scales using a subset of the items from Study 2. The “more profound” scale took the mean from the 5 items that were assigned the highest average profundity rating and the “less profound” scale consists of the 5 lowest scoring items (see Table 1). Both scales had acceptable internal consistency (Cronbach’s α = .81 and .75 for the relatively more and less profound items, respectively). As is evident from Table 2, the two scales performed very similarly. Indeed, heuristics and biases performance was significantly more strongly correlated with the scale that consists of the more profound items (r = –.36) than it was with the less profound items (r = –.23). The correlations were significantly different from one another according to a William’s test, t(187) = 2.08, p = .038, though both coefficients are significantly different from zero (p’s < .01).

Table 1:The 5 most and least profound bullshit statements in Pennycook et al.’s (2015) Study 2.

Table 2:Re-analysis of Pennycook et al.’s Study 2. Pearson product-moment correlations for 5 most profound and 5 least profound bullshit items. These data are for the full sample (N = 187). ^*** p < .001, ^** p < .01, ^* p < .05.

^*** p < .001

^** p < .01

^* p < .05.

This pattern of results is at variance with what we take to be the implications of Dalton’s argument. Specifically, if the observation that some participants may find transcendence in our bullshit items constrains our results, bullshit items that are more likely to be subjectively meaningful for participants should be less strongly negatively correlated (or even positively correlated) with analytic thinkingFootnote ¹. Our results indicate that, if anything, relatively more profound bullshit is more strongly negatively correlated with analytic thinking; perhaps because it is more difficult to detect that they are, in fact, bullshit.