Exploring different possible explanations for the null findings
Assuming that value was established via induction of hunger (see above), we did not find an effect of value of the unexpected object on the probability of its detection. This is consistent with Redlich et al., (
2019) who found no significant effect of short-term learned monetary value on inattentional blindness. Nevertheless, our findings seem surprising as (A) previous research has shown a clear effect of hunger on attentional bias towards food stimuli (Morris & Dolan,
2001; Piech et al.,
2010), (B) other studies have repeatedly shown that previously rewarded stimuli are preferentially processed and, thus, suggested that rewards are important in salience determination (Anderson, Laurent, & Yantis,
2011; Anderson & Yantis,
2012), and (C) noticing in an inattentional blindness paradigm has repeatedly been shown to be sensitive to other forms of value (attentional set: Most & Astur,
2005; Most et al.,
2001; Koivisto & Revonuso,
2007; self-related stimuli: Mack & Rock,
1998, or evolutionary predetermined value: New & German,
2015).
Different explanations for these null findings seem reasonable. (A) One explanation might be the existence of moderator variables. For example, previous research suggests that sex might determine specific food craving: some food stimuli rich in carbohydrates as ice cream have been found to be the most regularly craved foods among females (Christensen & Pettijohn,
2001). Based on this, Li et al. (
2015) already examined female participants with food-specific stimuli and food-specific cravings to investigate the effect of value on inattentional blindness. Potentially, cravings and, thus, the value of the food stimuli used in the present study were also higher for female than male participants as our food stimuli were rich in carbohydrates (chocolate, bread, and burger). To test this notion, we conducted a binary logistic regression analysis with the interaction term of sex and hunger condition as predictor and noticing of the unexpected food stimulus in the critical trial as dependent variable. Although we did not find a significant interaction effect for hunger condition and sex (
B = − 0.63, SE = 0.32, Wald = 3.96,
p = 0.071) on noticing, an additional chi-square test revealed that male participants were generally (i.e., independent of hunger condition) significantly more likely to notice unexpected food stimuli when their attention was engaged elsewhere (55%) than females (25%) [
χ2(1) = 8.61,
p = 0.003, RR(male/female) = 2.21 (95% CI: 1.29, 3.81),
BF10 = 32.51]. In contrast, no significant effects were found for the noticing rates of non-food stimuli between males (53%) and females (40%) [
χ2(1) = 1.08,
p = 0.299, RR(male/female) = 0.22 (95% CI: 0.85, 2.02),
BF10 = 0.53]. Despite these sex differences in regard to the detection of unexpected food stimuli, separate chi-square tests did not reveal significant effects of hunger for females [
χ2(1) = 0,
p = 1, RR(hungry/satiated) = 1 (95% CI: 0.38, 2.66),
BF10 = 0.30], nor for males [
χ2(1) = 7.17,
p = 1, RR(hungry/satiated) = 0.99 (95% CI: 0.62, 1.59),
BF10 = 0.32]. These exploratory results might function as an interesting starting point for future investigations.
(B) A further reason for our null findings might be the short time period we used to induce hunger and, thus, the value of food stimuli. Sixteen hours of food-deprivation can be easily achieved by missing out one meal as breakfast. Thus, differences in hunger state might be reflected on the VAS scales but might still not be strong enough to have practical implications for attentional orientation. However, this approach is commonly used in the field of hunger research (Evers et al.,
2011; Mogg et al.,
1998; Morris & Dolan,
2001) and previous research has shown attentional bias effects of hunger by even shorter experimental manipulations (6 h of fasting to implement hunger, Tapper, Pothos, & Lawrence,
2010). In contrast, studies focusing on more general aspects of cognition found equivocal results (see Benau, Orloff, Janke, Serpell, & Timko,
2014), which might demonstrate the complexity of short-term fasting on cognition. Fittingly, the general assumption that the concept of reward itself depends on a multitude of mechanism and determinants is supported by the literature on reward direction (gain vs. loss; Kahneman & Tversky,
1979) and the subjective reward value and probability (Chapman, Gallivan, & Enns,
2015). The manipulation chosen in the present study might be potent enough to have an impact on a sensitive measure as reaction times (Piech et al.,
2010; Tapper et al.,
2010), but not on the binary measure of awareness. Similar, such measure differences have also been found for priming effects (Kreitz et al.,
2015) and monetary value (Redlich et al.,
2019).
In contrast, other value-stimulus associations are based on long-term processes; meaningful and overlearned words (“Stop”, Mack & Rock,
1998) or threatening objects (spiders, New & German,
2015) have been found to successfully influence inattentional blindness. Potentially, including the personal food-craving trait into the analysis yields effects as it is also based on a long-term association process and might create a higher value for food stimuli. Consequently, we explored the individual food-craving trait as an additional variable that might modulate the relationship between hunger condition and noticing of the unexpected food stimulus in the critical trial. However, there was no significant interaction effect for hunger condition and individual food-craving trait on noticing [
B = 0.01, SE = 0.01, Wald = 0.33,
p = 0.567]. It seems that even general food craving as a trait is not strong enough to increase the food stimulus´ value and its likelihood to be noticed in an inattentional blindness paradigm.
(C) Another explanation for the null finding might be that in our study only 66% of the participants consciously perceived the shape of the unexpected stimulus when they said they noticed something in addition to the cross, whereas in contrast, everyone was able to choose the correct location. Potentially, the stimulus strength was not high enough for all participants to process the inherent meaning of the stimuli. We took this potential caveat into account and redefined noticing as “having noticed something in addition” and “being able to identify the correct shape”. However, there was no significant effect of hunger condition on the noticing of food stimuli [χ2(1) = 2.09, p = 0.148, RR(hungry/satiated) = 0.52 (95% CI: 0.22, 1.27), BF10 = 0.6] in this case, either.
(D) There might not be a general effect of hunger on the detection of food stimuli. That is, hunger might have an influence on attentional selection but this effect might not always be strong enough to be found in every paradigm. Previous studies showed that hunger leads to an increase in selective attention (Mogg et al.,
1998), improves the memory advantage for food stimuli (Morris & Dolan,
2001), and limits attentional shifting (Piech et al.,
2009). The only study so far showing direct effects of craving on the detection of unexpected food stimuli was the quasi-experimental study by Li et al. (
2015) that investigated effects of ice cream craving on noticing ice cream stimuli. Building on these results, we aimed to prepare the grounds for a general effect of value of the unexpected object on inattentional blindness through the comprehensive inclusion of different sexes, different food stimuli, and an overall food craving in our study. However, we did not find any effect of value based on our experimental hunger manipulation, indicating that the effects of Li and colleagues might not easily be generalized. Admittedly though, these results should be treated with caution; additional Bayesian analyses only moderately supported our null findings in contrast to the alternative hypothesis.
Prospects: the general impact of value on inattentional blindness
With regard to the general inattentional blindness literature, our findings lead to the assumption that effects of value cannot simply be generalized. This is in line with other types of value as faces, whose effects have been investigated in the phenomenon of inattentional blindness; several studies found that faces were more likely to be noticed compared to other stimuli (Devue, Laloyaux, Feyers, Theeuwes, & Brédart,
2009; Lee & Telch,
2008; Mack & Rock,
1998) and argued that this effect is driven by the stimuli’s importance (Mack, Pappas, Silverman, & Gay,
2002). In contrast, Mack and Clarke (
2012) showed that the presence of faces does not lead to higher noticing rates of an unexpected scene. Based on general assumptions about reward direction (gain vs. loss; Kahneman & Tversky,
1979), the subjective reward value (Chapman et al.,
2015), and our failed attempt to extend a stimulus-specific effect of semantic value on inattentional blindness towards a more general effect of semantic value, we theorize that the concept of semantic value can probably be divided into subtypes, based on the values’ characteristics and underlying mechanisms. Specifically, we propose that the semantic value of a specific stimulus or stimulus group is based on certain characteristics.
The first characteristic is the length of time during which an object is associated with semantic value (long term vs. short term), so that semantic value created through a long-term learning process might be stronger compared to semantic value created through a short-term learning process. Thus, the semantic value of spiders and snakes (New & German,
2015) can be seen a strong, since it is learned in an evolutionary long-term process, whereas the semantic value of stimuli associated with monetary reward might be seen as weak since this was learned in a period of only 20 min (Redlich et al.,
2019).
The second characteristic might be the quality of the association process, that is, whether it constitutes a high-relevance situation. Traumatic experiences might create a strong semantic value for stimuli associated with such a traumatic experience, whereas everyday experiences might not create a strong semantic value for stimuli associated with usual daily experiences.
The third characteristic could be the valence direction of the associated semantic value (positive vs. negative). For example, happy faces seem to be associated with stronger semantic value compared to frowning or sad faces (Lee & Telch,
2008; Mack & Rock,
1998).
The fourth characteristic that might influence the semantic value of a specific stimulus or stimulus group is the attentional set formed by context factors. The attentional set can be described as the “tuning” of one’s attention to prioritize certain features over others (Most,
2013) and, thus, strengthens the value of the prioritized features. Such “tuning” can be caused by environmental aspects; in a task we tune our attention to relevant stimuli that help us successfully perform this task, for example, triangular shapes or red stimuli in a computer task. In a traffic situation, our attention is more “tuned” to detect a human than to detect a kangaroo, as we might have experienced more men in business suits than kangaroos crossing a street in the city (Pammer & Blink,
2013).
We predict that a combination of these characteristics defines the semantic value of a specific object or event for a specific person in a specific context. Some characteristics, as the length of time during which a value association was learned, might apply to a large extent of the population, whereas others might only apply for a few. Therefore, we argue, that most value-driven attentional amplification might be sufficient to show in sensitive measures as reaction times (e.g., Mogg, et al.,
1998; Redlich et al.,
2019), but might not always suffice to help an object cross the threshold of awareness under conditions of inattention.