effects of benefits versus costs attended early
in a decision.REFERENCES AND NOTES- W. Kool, J. T. McGuire, Z. B. Rosen, M. M. Botvinick,J. Exp.
Psychol. Gen. 139 , 665–682 (2010). - A. Westbrook, D. Kester, T. S. Braver,PLOS ONE 8 , e68210 (2013).
- M. A. Apps, L. L. Grima, S. Manohar, M. Husain,Sci. Rep. 5 ,
16880 (2015). - J. D. Salamoneet al.,Behav. Processes 127 ,3–17 (2016).
- A. G. E. Collins, M. J. Frank,Psychol. Rev. 121 , 337– 366
(2014). - M. J. Frank, R. C. O’Reilly,Behav. Neurosci. 120 , 497– 517
(2006). - E. Aarts, M. van Holstein, R. Cools,Front. Psychol. 2 , 163 (2011).
- A. Westbrook, T. S. Braver,Neuron 89 , 695–710 (2016).
- N. D. Volkowet al.,Mol. Psychiatry 16 , 1147–1154 (2011).
- P. J. Cocker, J. G. Hosking, J. Benoit, C. A. Winstanley,
Neuropsychopharmacology 37 , 1825–1837 (2012). - R. Cools,Curr. Opin. Behav. Sci. 4 , 152–159 (2015).
- E. Aartset al.,Neuropsychologia 62 , 390–397 (2014).
- S. G. Manoharet al.,Curr. Biol. 25 , 1707–1716 (2015).
- M. H. M. Timmer, E. Aarts, R. A. J. Esselink, R. Cools,
Eur. J. Neurosci. 48 , 2374–2384 (2018). - S. McGuiganet al.,Brain 142 , 719–732 (2019).
16. M. I. Froböseet al.,J. Exp. Psychol. Gen. 147 , 1763–1781 (2018).
17. J. G. Hosking, S. B. Floresco, C. A. Winstanley,
Neuropsychopharmacology 40 , 1005–1015 (2015).
18. R. C. Spencer, D. M. Devilbiss, C. W. Berridge,Biol. Psychiatry
77 , 940–950 (2015).
19. R. Cools, M. D’Esposito,Biol. Psychiatry 69 , e113–e125 (2011).
20. P. Piray, H. E. M. den Ouden, M. E. van der Schaaf, I. Toni,
R. Cools,Cereb. Cortex 27 , 485–495 (2017).
21. L. Schmidt, M. Lebreton, M.-L. Cléry-Melin, J. Daunizeau,
M. Pessiglione,PLOS Biol. 10 , e1001266 (2012).
22. W. M. Pauli, R. C. O’Reilly, T. Yarkoni, T. D. Wager,Proc. Natl.
Acad. Sci. U.S.A. 113 , 1907–1912 (2016).
23.N.D.Volkowet al.,Am. J. Psychiatry 155 , 1325–1331 (1998).
24. N. D. Volkowet al.,J. Neurosci. 21 , RC121 (2001).
25. G. Jocham, T. A. Klein, M. Ullsperger,J. Neurosci. 31 ,
1606 – 1613 (2011).
26. C. Eiseneggeret al.,Neuropsychopharmacology 39 , 2366– 2375
(2014).
27. I. Krajbich, C. Armel, A. Rangel,Nat. Neurosci. 13 , 1292– 1298
(2010).
28. J. F. Cavanagh, T. V. Wiecki, A. Kochar, M. J. Frank,J. Exp.
Psychol. Gen. 143 , 1476–1488 (2014).
29. T. V. Wiecki, I. Sofer, M. J. Frank,Front. Neuroinform. 7 , 14
(2013).
30.ni~b 0 +b 1 (gBenA–gBenB)+b 2 (gCostA–gCostB)+b 3 gBenDVBen+
b 4 gCostDVCost+b 5 gCostDVBen+b 6 gBenDVCost(Eq. 1), where
versus costs, later gaze simply reflected a la-
tent choice.
Finally, we tested whether the effects of do-
pamine on choice could be attributed to these
dynamic decision processes. Indeed, both high-
er dopamine synthesis capacity [on placebo;
Eq. 1 ( 30 ): (b 3 + b 5 )/2; Pearson r = 0.30, P = 0.039;
Fig. 4D] and methylphenidate [tpaired,45 =
2.54, P = 0.015; Fig. 4E] increased the effect
of benefits on evidence accumulation. The cor-
responding effect of sulpiride on cost was not
significant [Eq. 1 ( 30 ): (b 4 + b 6 )/2; tpaired,45 =
- 1.41; P = 0.17; see the supplementary dis-
cussion]. We further found that methylpheni-
date amplified the effects of benefits on drift
rate even when only modeling pre-bifurcation
gaze [tpaired,45 = 2.44; P = 0.019) before the
latent choice. Collectively, our results sup-
port that striatal dopamine enhances moti-
vation for cognitive effort by amplifying the
SCIENCE 20 MARCH 2020•VOL 367 ISSUE 6484^1365
Fig. 4. Gaze dynamically biases and then reflects implicit choice.(A)Gaze
attribute model. Early gaze amplified the effect of attended versus unattended
attributes on choice during evidence accumulation to a decision threshold (a). Late
gaze reflected the to-be-selected response. (BandC) Model simulations (red)
predicted choice (gray) (B, split by median proportion gaze at the higher value offer)
and reaction time (C) distributions. (D) Benefits effect on drift rate correlate with
dopamine synthesis capacity (95% CI shown). (E) Methylphenidate enhances the
benefit effect. (FandG) Posterior parameter densities from models fit alternately
with pre- or postbifurcation gaze on placebo. (F) Additive benefit (b 1 =–0.030;
P= 0.076) and cost (b 2 = 0.020;P= 0.81) gaze terms were approximately zero
before bifurcation and reliably positive after bifurcation (b 1 = 0.10;P< 2.2 × 10–^16
andb 2 = 0.11;P= 0.0031). (G) Multiplicative interaction terms reveal that the effects
of benefits (b 3 – b 5 = 0.40;P= 0.0024) and costs (at trend level;b 4 – b 6 = 0.12;
P= 0.060) were larger when fixating the respective attribute before bifurcation,
whereas neither term was different from zero after bifurcation (b 3 – b 5 = 0.07;P= 0.27
andb 4 – b 6 =–0.060;P= 0.70). Error bars indicate ± SEM.RESEARCH | REPORTS