No evidence for a common self-bias across cognitive domains

Abstract

It is generally acknowledged that humans have an egocentric bias; processing self-related stimuli in a specialised, preferential manner. The self-bias has been studied within cognitive domains such as memory, attention and perception; but never across cognitive domains in order to assess whether self-biases are a product of a common bias, or independent. This has relevance for conditions such as Autism Spectrum Disorder: certain self-biases are reduced in those with autism, but the pattern of results is not consistent across different cognitive domains. Self-bias was measured across the attentional and perceptual domains on two well-established tasks: the attentional blink (attention) and shape-label matching (perception) tasks. Processing of each participant's own name was compared to processing of the name of another individual very familiar to the participant (to control for familiarity), and the name of an unfamiliar other. In the attentional domain, the attentional blink for the participant's own name was reduced compared to that for the name of a familiar or unfamiliar other. In the perceptual domain, participants showed stronger associations between their own name and a geometric shape than between the other classes of names and associated shapes. Thus, strong evidence of a self-bias, independent of familiarity, was found on both tasks. However, across two experiments, the magnitude of the self-bias on the attentional blink and shape-label matching tasks was not correlated, supporting the idea that self-biases across cognitive domains are distinct. Furthermore, in contrast with extant models, neither type of self-bias was predicted by autistic traits.

Introduction

Over the centuries, the concept of ‘the self’ has fascinated philosophers and psychologists alike. Despite the lack of consensus on what exactly constitutes the human self-concept (for two recent views, see Gallagher, 2013; Northoff, 2016), it is generally agreed that our experience of a unified sense of self strongly biases the way we process the world around us. Across different sensory modalities, and while performing a multitude of tasks, humans show a strong bias for processing self-related information such as one's own face (Keyes & Brady, 2010) or name (Wood & Cowan, 1995). This self- (or egocentric) bias is argued to be closely linked to social functioning: a strong bias to process self-relevant cues is likely to lead to a more accurate model of the self and of those similar to us, enabling us to better infer their emotions and thoughts in a given situation (Conway, Catmur, & Bird, 2019; Goldman, 2006; Mitchell, 2009; Nijhof & Bird, 2019). However, should the self-bias be so extreme that it results in too little processing of other-related information, the development of associative links between cues to particular states in others and corresponding states in the self may be impacted. These links have been argued to be crucial in the development of imitation (Catmur, Walsh, & Heyes, 2009; Cook, Bird, Catmur, Press, & Heyes, 2014), empathy (Bird & Viding, 2014; Heyes & Bird, 2007; Quattrocki & Friston, 2014), and Theory of Mind (Ondobaka, Kilner, & Friston, 2017). Such an extreme focus on the self (‘egocentrism’) has been argued to be a cardinal feature of Autism Spectrum Disorder (ASD; Frith & de Vignemont, 2005; Lombardo & Baron-Cohen, 2010), although several studies have since found the opposite pattern, that is: a decrease in specific aspects of self-preferential processing in ASD (Uddin, 2011; Williams, 2010). Studying the self-bias may therefore lead not only to understanding how individuals process self-related information and develop a self-concept, but may also shed light on their socio-cognitive ability and how this may be impacted in conditions such as ASD.

While a self-bias has been observed on a number of tasks, the exact cognitive process it impacts is not yet clear. Across tasks, it has been claimed that a self-bias can be observed for several cognitive domains, including perception, memory and attention (see Cunningham & Turk, 2017, for an overview). For example, evidence for the claim that self-bias can be observed in the perceptual domain is provided by work by Sui and colleagues using the ‘shape-label matching task’ (Humphreys & Sui, 2015; Sui, He, & Humphreys, 2012). In this task participants learn to associate the self or others with neutral geometric shapes. When they subsequently have to judge whether specific shape-label combinations are correct, they show faster and more accurate performance for associations made to the self. Evidence that this effect is on perception (rather than attention) is provided by the fact that 1) the effect is maintained even when the self- or other-label is cued by the associated shape (minimising attentional effects) (Humphreys & Sui, 2016; Wang, Humphreys, & Sui, 2016), 2) functional Magnetic Resonance Imaging indicated less engagement of the frontoparietal attentional network for self-associations than other associations (Sui, Rotshtein, & Humphreys, 2013), and 3) the self-bias influences reaction times in the same manner as manipulations of perceptual salience such as increasing or decreasing contrast (Sui, Liu, Mevorach, & Humphreys, 2013).

In the attentional domain, Shapiro, Caldwell, and Sorensen (1997) have investigated the bias for one's own name in the context of the attentional blink, using rapid serial visual presentation (RSVP): presenting streams of visual stimuli at a fast, fixed rate. The attentional blink describes the phenomenon that between approximately 200 and 500 ms after the presentation of a first target (T1) in a RSVP stream, the ability to detect a second target (T2) is significantly reduced (Raymond, Shapiro, & Arnell, 1992). By presenting the T2 at different temporal lags with respect to T1, the time course of the attentional blink can be studied with great precision. Theoretical accounts of the attentional blink suggest there is a temporal capacity-limit on selective attentional resources, and therefore attentional demands of processing the first target prevent these resources from being applied to the second target, thus creating the attentional blink (Dux & Marois, 2009; Martens & Wyble, 2010). This implies that people will show a reduced attentional blink for highly salient stimuli, as these require less attentional resources to be processed. Assuming the own name is such a highly salient stimulus, Shapiro et al. (1997) hypothesised that the attentional blink would be reduced for one's own name (compared to another name). In a series of experiments, they tested attentional blink effects, contrasting the participant's own name with another person's name. It was indeed found that when presented as T2, one's own name, in contrast to a stranger's name, is resistant to the attentional blink: detection of one's own name was similarly high at all temporal lags. Since the attentional blink is thought to reflect the depletion of attentional resources, the authors concluded that the own name survives the attentional blink because less resources are required to process it (see also Tibboel, De Houwer, Van Bockstaele, & Verschuere, 2013).

Finally, in the domain of memory, it has been repeatedly shown that humans have superior memory for material that has been related to the self, a finding referred to as the self-reference effect (Symons & Johnson, 1997). The self-reference effect was initially found for stimuli that allow elaborate self-related processing, such as for personality traits after participants were asked to judge whether a trait applies to themselves, as compared to a specific other. However, a self-reference effect is also present for random objects after it is stated that they are owned by the participant versus by someone else: the ‘ownership effect’ (Cunningham, Turk, Macdonald, & Macrae, 2008).

A key question to be answered with respect to the effect of the self-bias on perception, attention and memory is whether these effects are 1) really one effect, 2) distinct but related, or 3) distinct and unrelated. The different self-biases could be considered one effect (1) if, for example, the effects on perception and memory were the product of the attentional self-bias such that self-related shapes were perceived faster/as if they had greater contrast, or items/traits remembered better, as a result of being accorded greater attention. Alternatively, the self-biases observed in perception, attention and memory could be distinct effects, but driven by a common self-bias (2) such that, across individuals, the magnitude of the self-bias in one domain predicts the magnitude of the self-bias in another domain. Alternatively, the self-biases may be distinct effects and the magnitude of each effect unrelated to the others (3) such that, across individuals, there is no correlation between the magnitude of different self-biases. It is this question that the current study attempts to address.

Specifically, the current study attempts to determine whether the self-biases reported in the perceptual and attentional domains are distinct by asking participants to complete both the shape-label matching task and the attentional blink task. The perceptual basis of the self-bias observed in the shape-label matching task is supported by the psychological and neurological evidence indicated previously (Humphreys & Sui, 2016; Sui, Liu, et al., 2013; Sui, Rotshtein, & Humphreys, 2013). However, problematically, the attentional nature of the self-bias observed in the attentional blink task is called into question by the results of Shapiro et al. (1997) when comparing the effects of using one's own name (in contrast to a stranger's name) as T1 in the attentional blink paradigm. If the attentional blink is driven by a temporal capacity-limit on selective attentional resources such that attentional demands of processing the first target prevent these resources from being applied to the second target, and one's own name is less subject to the attentional blink when shown as T2 as it requires less attentional resources to be processed, then one would expect that one's own name should elicit less of an attentional blink if used as T1 in comparison to a stranger's name. However, Shapiro et al. (1997) did not find any difference in the attentional blink elicited following one's own name, in comparison to a stranger's name, when used as T1. Importantly however, sample size for this experiment was extremely low (N = 8), and we are not aware of any attempt to replicate this null result.

Accordingly, Experiment 1 attempted to replicate this experiment with a larger sample size in order to establish whether any evidence for an attentional basis of the self-bias observed on the attentional blink could be found. If truly attentional, when one's own name is used as T1 the attentional blink elicited should be reduced in magnitude compared to other names, as less attentional resources are required to process it, leaving additional resources for T2. In addition, we also sought to control for effects of familiarity (to the extent that this is possible for one's own name). Problematically for many studies of the self-bias, self-related stimuli tend to be more familiar than other-related stimuli, making it ambiguous as to whether any effects are a product of self-relatedness specifically, or are instead a general effect of stimulus familiarity. To reduce this ambiguity, studies of self-bias typically contrast self-related stimuli (e.g. one's own name) with familiar but not self-related stimuli (e.g. the name of someone close to you). When compared to unfamiliar stimuli (e.g. a stranger's name), distinct effects of self-relatedness and familiarity can be observed (Cygan, Tacikowski, Ostaszewski, Chojnicka, & Nowicka, 2014; Nijhof, Goris, Brass, Dhar, & Wiersema, 2018; Tacikowski et al., 2011; Yang, Wang, Gu, Gao, & Zhao, 2013). To date, studies of the self-bias using the attentional blink have not used a familiarity control, so it was included in both Experiments 1 and 2.

In order to address the relationship between self-biases observed in the shape-label matching task and the attentional blink task, Experiment 2A compared the magnitude of these self-biases within the same individuals. Experiment 2B constituted a replication of 2A and also investigated the reliability of self-bias effects within the same task over a one week period. Surprisingly, to our knowledge, self-biases have never been compared in the same individuals before, and therefore there are no relevant data on the relationship between different self-biases. At the theoretical level however, the relationship between self-biases has attracted a lot of attention with respect to the differences in self-related processing observed in those with ASD. This theoretical work has been based on separate studies comparing the magnitude of different self-biases in autistic¹ and neurotypical individuals (Nijhof & Bird, 2019). For example, ownership self-reference effects have been shown to be diminished in autistic individuals (Grisdale, Lind, Eacott, & Williams, 2014), and responses to one's own name and face are reduced in autistic children and adults (Cygan et al., 2014; Nijhof et al., 2018; Werner, Dawson, Osterling, & Dinno, 2000). In contrast, the size of the self-bias observed on trait memory tasks and on the shape-label matching task did not differ between autistic and neurotypical individuals (Lind, Williams, Nicholson, Grainger, & Carruthers, 2019; Williams, Nicholson, & Grainger, 2017). Results such as these have led researchers to argue for a distinction between the representation of psychological and physical aspects of the self (Uddin, 2011; Williams, 2010), and a distinction between first- and second-order representations of the self (Williams et al., 2017) in autism. As noted, however, these theoretical distinctions are based on empirical work which compares self-biases in different samples of individuals rather than in the same individuals, further highlighting the need to compare self-biases in the same individuals. Given the relevance to ASD, autistic traits were measured in both Experiments 1 and 2.