Motto: "Do you see yonder cloud that's almost in shape of a camel? Polonius: By the mass, and 'tis like a camel, indeed. Hamlet: Methinks it is like a weasel. Polonius: It is backed like a weasel. Hamlet: Or like a whale? Polonius: Very like a whale." --William Shakespeare, Hamlet 1. Introduction
The famous dialogue between Hamlet and Polonius nicely illustrates a typical way how perceptual reports are influenced by non-perceptual cognitive mechanisms. However, to know whether somebody indeed experiences something after having been biased by a suggestible talk or whether this is just a biased responding without any change in the responder's direct perception requires psychometric measurement of suggestibility as well as direct perception. Adherents to the predictive coding theory have listed many experimentally verified factors such as contextual expectations, priming, personality traits, neuropsychiatric vulnerabilities etc. that can predispose people to non-veridical perception (e.g., Clark 2013, de Lange et al. 2018; Fletcher and Frith 2009, Friston 2010, Palmer et al. 2017, Teufel and Fletcher 2020). The outcome of the perceptual process depends on the relative contribution of (a) prior information encoded in the cognitive system and (b) actual direct sensory evidence (de Lange et al. 2018, Fleming 2020, Gilbert and Li 2013, Sterzer et al. 2008, Teufel and Fletcher 2020). If the former overweighs the latter, experience is close to what was expected. In most cases with mentally healthy subjects, good enough perceiving conditions, typical circumstances and contexts, the expected and the real coincide. However, strong priors can also lead to non-veridical experience when the circumstances allow.
The sources of expectation-based bias can be categorized as higher or lower level factors capable of forming subjective perception (Clark 2013, Corlett et al. 2019, Fleming 2020, Lupyan 2015, Series and Seitz 2013, Summerfield and de Lange 2014, Tulver et al. 2019). Higher levels include beliefs, general conceptual knowledge stored in the cognitive system, probabilistic estimations etc. Typicality of situations also forms the learned contextual priors capable of misperception bias for example when real situations are not typical. Lower levels include structural constraints 'hardwired' in the perceptual system such that even higher-level knowledge speaking against the way the habitat is perceived cannot change the percept (e.g. illusory contours or motion extrapolation). The effects of cognitive and perceptual level priors on non-veridical perception have been well studied (e.g. Balcetis and Dunning 2006, de Lange et al. 2018, Gilbert and Li 2013, Lupyan 2017, O'Callaghan et al. 2017), including when mentally normal perceivers experience objects that are not actually presented (Aru and Bachmann 2017, Aru et al. 2018, Partos et al. 2016). We will call the latter cases 'normal hallucinations' (to distinguish from 'real hallucinations' for difference from clinical connotations). Effects of schizotypy and autism on hallucination proneness in (pre)clinical populations have been widely studied, with no full consensus on the precise mechanisms achieved as yet (Behrendt and Young 2004, Fletcher and Frith 2009, Karvelis et al. 2018, Lawson et l. 2014, Palmer et al. 2017, Pellicano and Burr 2012, Powers et al. 2017, Reed et al. 2008, van de Cruys et al. 2014, 2017). However, there is much less research on the higher-level effects of stable personality traits of neurotypical population on the propensity to experience actually formed expectation-based normal hallucinations. One of the aims of this exploratory study is to contribute to this shortcoming.
The other aim stems from the fact that in the majority of cases the expectation based misperception by normal subjects has been demonstrated in the experimental paradigms where either interpretation of ambiguous real objects was examined or real stimuli in the form of pure noise were used as the 'mold' from where illusory percepts emerged (e.g. Balcetis and Dunning 2006, Bruger et al. 1993, de Lange et al. 2018, Gilbert and Li 2013, Jakes and Hemsley 1986, Lupyan 2017, O'Callaghan et al. 2017, Partos et al. 2016, Rieth et al. 2011). Normal hallucinations in the conditions where they emerge on an empty background (Aru and Bachmann 2017, Aru et al. 2018, Vetik et al. 2020) have not been used in the context of the effects of higher-level personality related priors. We combine the above described two aims in a single study.
Looking for a more or less stable personality characteristic that is likely to influence normal hallucinations understandably leads to considering suggestibility. (For more information on this construct or trait see Gudjonsson 1984, 1992, 1997 and Terhune et al. 2017.) Suggestibility as a measurable trait almost by definition refers to how much a suggestion given by somebody is capable of changing someone else's behavior--for example behavior explicated in reporting own cognitive-perceptual experience after having been subjected to formation of perceptual expectations. There are direct and indirect suggestive influences of which both are typically applied by some means of instructions and questions (Pohl 2016, Polczyk and Pasek 2006). Suggestibility as assessed by Gudjonsson's GSS2 (1992, 1997) belongs to the indirect variety of suggestibility. It has been argued that suggestibility and suggestions positively correlate with hallucination proneness (Alganami et al. 2017, McGeown et al. 2012, Young et al. 1987). On the other hand, there is data allowing to doubt this association (e.g. Merckelbach and van de Ven 2001, Smith and Gudjonsson 1995). Moreover, even though the positive effect of suggestibility on hallucinations appears to be real, in many accounts it can be explained by reporting biases rather than change in vividness of direct experience (Alganami et al. 2017, Merckelbach and van de Ven 2001, Partos et al. 2016). We will examine how much the subjective clarity of the normal hallucinations which are based on experimentally formed expectation are related to Gudjonsson type suggestibility. We hypothesize that perceivers who rate their hallucinations as clearer have also higher level of suggestibility in general.
For our purposes, we used two instruments. First, for assessing the level of suggestibility, Gudjonsson's scale GSS2 (Gudjonsson 1997) was adopted. Second, for obtaining and measurement of normal hallucinations the method used earlier by Aru, Tulver and Bachmann (2018) was implemented.
Gudjonsson Suggestibility Scale 2 (GSS2; Gudjonsson 1997) is based on a narrative paragraph describing a young boy who loses control of his bike. This narrative is read to the tested persons who are thereafter asked to report all that can be recalled about the story. 20 specific questions are asked, 15 of which are misleading and suggestive. After answering the questions, the person is told that (s)he made a number of errors (independent of whether errors have been made), and thus it is necessary to ask all the questions again. The scale allows four scores: (1) Yield 1--the extent to which the tested give in to misleading questions. The range of possible scores is 0-15. (2) Yield 2--the extent to which they give in to misleading questions after negative feedback (interrogative pressure). The range of possible...