## Correctness of recognition I analysed whether pronounceability can influence not only the tendency to answer that the presented word had been previously seen, but also correctness of the answers (i.e., correct recognition). Therefore, I included interactions between previous presentation and both pronounceability factors in the model. The results suggested that pronounceability influenced correct recognition; however, I found only a linear effect of pronounceability, not quadratic. In particular, the effect of previous presentation was larger for more easily pronounceable words as indicated by the interaction between previous presentation and pronounceability, z = 2.83, p = .005, ratio of OR = 1.12, 95% CI = [1.04, 1.22]. This can be seen as the increasing spread between the dotted and dashed regression curves in the figure below. Correctly judging whether a word had been previously presented seemed to be easier for easy-to-pronounce words. The interaction between previous presentation and pronounceability squared was not significant, z = 0.26, p = .79, ratio of OR = 1.01, 95% CI = [0.95, 1.07]. The lack of the interaction suggests that the quadratic effect of pronounceability does not influence correctness of the answer. Rather, it only biases participants to say less often that they had seen the easy- and hard-to-pronounce words. On the other hand, more easily pronounceable words are more likely to be correctly recognized, but there is no linear effect of pronounceability that would influence the tendency to answer that the word had been shown before. ![correctness of recognition results][1] The figure depicts the association between pronounceability and recognition at the item level. The regression curves were computed using polynomial regression without any other predictors and with the proportion of “seen” answers as the dependent variable. The curve in full line corresponds to regression computed with all trials, the dotted curve presents regression only for trials where the word had been previously presented, and the dashed curve presents regression only for trials where the word had not been previously presented. ## Dominant hand ### Liking Apart from the above mentioned predictors common to all analyses, I included in the model whether the “like” answer corresponded to a key on participant’s dominant hand side. The effect of the answer on participant’s dominant hand side was not significant, z = 0.01, p = .99, OR = 1.00, 95% CI = [0.91, 1.10]. ### Recognition Participants were somewhat less likely to answer that they had seen the word if this answer was on their dominant hand side, z = -1.92, p = .06, OR = 0.91, 95% CI = [0.83, 1.00]. ### Liking decision time While the effect of dominant hand use answering the question seemed to differ between participants, it did not influence liking decision times in general, t(194.1) = -0.51, p = .61, b = -0.003, 95% CI = [-0.016, 0.009]. The effect of the answer on the side of the dominant hand was not significant for a simpler model and was removed from the analysis due to convergence issues occurring in model estimation. ### Recognition decision time Decision times were influenced neither by use of dominant hand in answering, t(6872.8) = 1.17, p = .24, b = 0.008, 95% CI = [-0.005, 0.021], nor by the key associated with the “seen” answer, t(6874.5) = 0.89, p = .37, b = 0.006, 95% CI = [-0.007, 0.019]. ### Discussion I did not find that participants were faster responding using their dominant hand. While a previous study (de la Vega, de Filippis, Lachmair, Dudschig, & Kaup, 2012) showed that people are faster to classify words as positive with dominant hand and as negative with non-dominant hand, I found no effect of the answer on participant’s dominant hand side on speed of the answer. It is possible that the difference in results was due to differences in the tasks. Whereas in the study by de la Vega et al. (2012), participants only classified words as positive or negative, in the present study participants had to judge their liking of words. Furthermore, de la Vega et al. (2012) argue that the effect of response side on reaction times occurs only if the mapping of valence to a side is made salient, which was not the case in the present experiment. ### References de la Vega, I., De Filippis, M., Lachmair, M., Dudschig, C., & Kaup, B. (2012). Emotional valence and physical space: Limits of interaction. *Journal of Experimental Psychology: Human Perception and Performance, 38*, 375-385. <br><br> ## Presentation order The presentation order was recoded to a -0.5 to 0.5 scale. ### Reading time The time needed to read a word was longer in later trials, t(176.3) = 3.87, p < .001, b = 0.081, 95% CI = [0.040, 0.122]. ### Liking The effect of order was not significant, z = -0.55, p = .58, OR = 0.95, 95% CI = [0.78, 1.15]. ### Recognition Participants were more likely to say that they had not seen the presented word in later trials, z = -3.10, p = .002, OR = 0.75, 95% CI = [0.62, 0.90], which may be a result of memory decay or interference by the previously presented words. ### Liking decision time Liking decision times were faster in later trials, t(161.1) = -4.01, p < .001, b = -0.057, 95% CI = [-0.084, -0.029]. Recognition decision time Contrary to decision times for liking, later trials were associated with slower decision times for recognition, t(138.1) = 2.95, p = .004, b = 0.040, 95% CI = [0.013, 0.067]. [1]: https://mfr.osf.io/export?url=https://osf.io/rbpq9/?action=download%26mode=render%26direct%26public_file=True&initialWidth=848&childId=mfrIframe&parentTitle=OSF+%7C+Fig3b.png&parentUrl=https://osf.io/rbpq9/&format=720x720.jpeg