Results and Discussion ---------------------- **STUDY 1** **Subjects** The subjects were 120 students from the University of Warwick (72.5% female, *M* = 19.24, *SD* = 1.26, range 18-22 years). Subjects received course credit for an introductory psychology course (56.7%) or £3 (43.3%) for their participation. We checked for differences between subjects who received credit and those who were paid, and found that paid subjects were slightly older than credit subjects *t*(92.62) = -7.15, *p* < .001, *r* = .60, (*M* paid = 20.06, *SD* payed = .12, *M* course credit = 18.63, *SD* course credit = .11) and paid subjects were less confident in their identifications *t*(118) = 2.52, *p* < .05, *r* = .23 (*M paid* = 4.69, *SD* payed = .1.42, *M* course credit = 5.28, *SD* course credit = 1.13). However, the proportion of paid versus credit subjects allocated to each condition was similar *χ* 2 (1) = .54, *p*(two-tailed) = .58, Cramer’s V = .07 so these differences should not have affected our results. **Results** *Recognition accuracy* We conducted a Chi-Square comparing the proportion of correct identifications in the experimental and the control conditions. Recognition accuracy was similar regardless of whether subjects verbalized the target face or not. Specifically, 56% of verbalization subjects and 65% of control subjects correctly identified the target, *χ* 2 (1) = .87, *p*(one tailed) = .23, Cramer’s V = .09. We compared the ratio of the percentage of subjects who made a misidentification to the percentage who indicated “not present” across the two conditions. Verbalization subjects were more likely to make misidentifications than control subjects when they made an error. Specifically, 58% of errors in the verbalization condition were misidentifications compared to 19% of errors in the control condition, *χ* 2 (1) = 7.20, *p*(two-tailed) = .02, Cramer’s V = .39. *Confidence ratings* A 2 (Condition) x 2 (Correct vs. incorrect/miss) ANOVA on subjects’ confidence ratings revealed that subjects were significantly more confident in correct responses compared to incorrect responses (*M* correct = 5.42, *SD* correct = 1.17, *M* incorrect = 4.40, *SD* incorrect = 1.25, *F*(1,116) = 20.50, *p* < .000, ηp2 = .15). We found no significant difference between the mean confidence of subjects in the verbalization and control conditions (*M* verb = 5.00, *SD* verb = 1.29, *M* control = 5.05, *SD* control = 1.21) *F*(1,116) = .05, *p* = .82, ηp2 = .00, and no significant interaction between accuracy and condition *F*(1,116) = .113, *p* = .74, ηp2 = .001. ---------- **STUDY 2** **Subjects** The subjects were 120 students from the University of Warwick (72% female, *M* = 20.05, *SD* = 1.62, range 18-25 years). Subjects received £3 for their participation. **Results** *Recognition accuracy* We conducted a Chi-Square comparing the proportion of correct identifications in the experimental and control conditions. Recognition accuracy was marginally worse when subjects verbalized the target face. Specifically, 43% of verbalization subjects correctly identified the target compared to 60% of control subjects, *χ* 2 (1) = 3.34, *p*(one tailed) = .050, Cramer’s V = .17. We compared the ratio of the percentage of subjects who made a misidentification to the percentage who indicated “not present” across the two conditions. The number of misidentifications was similar regardless of whether subjects verbalized the target face or not. Specifically, 56% of errors in the verbalization condition and 79% of errors in the control condition, were misidentifications, *χ* 2 (1) = 3.38, *p*(two-tailed) = .119, Cramer’s V = .24. *Confidence ratings* A 2 (condition: verbalisation vs. control) x 2 (response: correct vs. incorrect/miss) ANOVA on subjects’ confidence ratings revealed that subjects were significantly more confident in correct responses compared to incorrect responses (*M* correct = 5.24, *SD* correct = 1.14, *M* incorrect = 4.60, *SD* incorrect = 1.27, *F*(1,116) = 7.80, *p* = .006, ηp2 = .063). We found no significant difference between the mean confidence of subjects in the verbalization and control conditions (*M* verb = 4.85, *SD* verb = 1.18, *M* control = 5.02, *SD* control = 1.31), *F*(1,116) = 0.07, *p* = .786, ηp2 = .001, and no significant interaction between accuracy and condition, *F*(1,116) = 0.01, *p* = .914, ηp2 = .000.