Reaching Out for Food: How Food Incentives Modulate Peripersonal Space Perception

Two experiments were conducted to determine, first, whether food items influence participants’ estimations of the size of their subjective peripersonal space. It was of particular interest whether this representation is influenced by satiated/hungry states and is differentially affected by valence and calorie content of depicted stimuli. Second, event-related brain potentials (ERPs) were used, in order to obtain information about the time course of the observed effects and how they depend on the spatial location of the food pictures. For that purpose, participants had to decide whether food items shown at various distances along a horizontal plane in front of them, were reachable or not. In Experiment 1, when participants were hungry, they perceived an increase of their peripersonal space modulated by high-calorie items which were experienced as being more reachable than low-calorie items. In Experiment 2, the reachability findings were replicated and early and late components of ERPs showed an attentional enhancement in far space for food items when participants were hungry. These findings suggest that participants’ subjective peripersonal space increased while being hungry, especially for high-calorie contents. Attention also seems to be oriented more strongly to far space items due to their expected incentive-related salience, expanding the subjective representation of peripersonal space.


Reachability -Experiment 1
Note: We tried first to fit the model with the crossed random effects of participants and Items and estimate the random slopes of condition interacting with position, per subject and per item with this structure: (1+Condition_code*Position_cent|ID) + (1+Condition_code*Position_cent|Items). However, this model failed to converge. Barr et al. (2013) suggest to stepwise down the complexity of the random effects until the model converges. Thus, to fit the model, we removed the interaction and removed one random slope: (1+Condition_code|ID)+(1|Items). Then, the model converged.

Reachability -Experiment 2
Note: We tried first to fit the model with the crossed random effects of participants and Items and estimate the random slopes of condition interacting with position, per subject and per item with this structure: (1+Condition_code*Position_cent | ID) + (1+Condition_code*Position_cent | Items). However, this model failed to converge. Barr et al. (2013) suggest to stepwise the complexity of the random effects until the model converges. As for Experiment 1, we again followed the step wise procedure suggested by Barr et al. (2013). To fit the model, we removed the interaction (Condition*Position), and we tried with (1+Condition_code|ID)+(1+ Condition_code | Items). Then, the model converged.   Table 1 Estimated parameters from both Experiment 1 and 2 of logistic mixed-effects regression models (binomial responses). Sjplot package -tab _model(), Lüdecke D (2020) τ 00 =Intercepts variance, τ 11 =Slopes variance, ρ01=Correlation between pairs (Intercepts -Slopes), σ 2= Residual error Note: For generalized linear models, the output is slightly adapted. Instead of Estimates, the column is named Log-Odds.

Linear mixed model (LMM)
For the Subjective food preference ratings part Linear Mixed Model (LMM) was performed. To fit an LMM in the lme4 package, we used the lmer() function (linear mixed effects regression).
Rating y si is the dependant variable Crossed random effects: We used these fixed effect terms (X) In lme4 the formula was:  Table 2 Estimated parameters from both Experiment 1 and 2 of Linear Mixed Models.

ERP -Analysis
For the ERP part, a Linear Mixed Model (LMM) was performed. To fit an LMM in the lme4 package, we used the lmer() function (linear mixed effects regression).

ERP -P1
Note: We tried first to fit the model with the crossed random effects of participants and Items and estimate the random slopes of condition interacting with distance, per subject and per item with this structure: (1+Condition_code*Distance_code|ID) + (1+Condition_code*Distance_code | Items). However, this model failed to converge. Barr et al. (2013) suggest to stepwise down the complexity of the random effects until the model converges. Thus, to fit the model, we removed the interaction and removed one random slope: (1+Condition_code|ID)+(1|Items). Then, the model converged.

ERP -LPC
Note: We tried first to fit the model with the crossed random effects of participants and Items and estimate the random slopes of condition interacting with distance, per subject and per item with this structure: (1+Condition_code*Distance_code | ID) + (1+Condition_code*Distance_code | Items). However, this model failed to converge. Barr et al. (2013) suggest to stepwise the complexity of the random effects until the model converges. As for the P1 mixed effect analysis, we again followed the step wise procedure suggested by Barr et al. (2013). To fit the model, we removed the interaction (Condition_code*Distance_code), and we tried with (1+Condition_code|ID)+(1+ Condition_code|Items). Then, the model converged.