科技科学

The neural basis of flavour–nutrient learning

Nature查看原文 ↗

Data availability

The data used to generate the main findings of this study have been publicly deposited at Dryad74 (https://doi.org/10.5061/dryad.h9w0vt4x2).

Code availability

No custom software or original algorithms were developed for use in this study. Any of the findings can be replicated using established software and the deposited data.

References

  1. Sclafani, A. Learned controls of ingestive behaviour. Appetite 29, 153–158 (1997).

    Article  CAS  PubMed  Google Scholar 

  2. Capaldi, E. D. Conditioned food preferences. In Psychology of Learning and Motivation (ed. Medin, D. L.) Vol. 28, 1–33 (Academic, 1992).

  3. Sclafani, A. Conditioned food preferences. Bull. Psychon. Soc. 29, 256–260 (1991).

    Article  Google Scholar 

  4. Holman, G. L. Intragastric reinforcement effect. J. Comp. Physiol. Psychol. 69, 432–441 (1969).

    Article  CAS  PubMed  Google Scholar 

  5. Berke, J. D. What does dopamine mean? Nat. Neurosci. 21, 787–793 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  6. Saunders, B. T., Richard, J. M., Margolis, E. B. & Janak, P. H. Dopamine neurons create Pavlovian conditioned stimuli with circuit-defined motivational properties. Nat. Neurosci. 21, 1072–1083 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Schultz, W. Multiple dopamine functions at different time courses. Annu. Rev. Neurosci. 30, 259–288 (2007).

    Article  CAS  PubMed  Google Scholar 

  8. Azzara, A. V., Bodnar, R. J., Delamater, A. R. & Sclafani, A. D1 but not D2 dopamine receptor antagonism blocks the acquisition of a flavor preference conditioned by intragastric carbohydrate infusions. Pharmacol. Biochem. Behav. 68, 709–720 (2001).

    Article  CAS  PubMed  Google Scholar 

  9. Touzani, K., Bodnar, R. J. & Sclafani, A. Acquisition of glucose-conditioned flavor preference requires the activation of dopamine D1-like receptors within the medial prefrontal cortex in rats. Neurobiol. Learn. Mem. 94, 214–219 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Touzani, K., Bodnar, R. J. & Sclafani, A. Activation of dopamine D1-like receptors in nucleus accumbens is critical for the acquisition, but not the expression, of nutrient-conditioned flavor preferences in rats. Eur. J. Neurosci. 27, 1525–1533 (2008).

    Article  PubMed  Google Scholar 

  11. Touzani, K., Bodnar, R. J. & Sclafani, A. Dopamine D1-like receptor antagonism in amygdala impairs the acquisition of glucose-conditioned flavor preference in rats. Eur. J. Neurosci. 30, 289–298 (2009).

    Article  PubMed  PubMed Central  Google Scholar 

  12. Zukerman, S., Ackroff, K. & Sclafani, A. Post-oral glucose stimulation of intake and conditioned flavor preference in C57BL/6J mice: a concentration–response study. Physiol. Behav. 109, 33–41 (2013).

    Article  CAS  PubMed  Google Scholar 

  13. Zimmerman, C. A. et al. A neural mechanism for learning from delayed postingestive feedback. Nature 642, 700–709 (2025).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Garcia, J., Ervin, F. R. & Koelling, R. A. Learning with prolonged delay of reinforcement. Psychon. Sci. 5, 121–122 (1966).

    Article  Google Scholar 

  15. Garcia, J., Lasiter, P. S., Bermudez-Rattoni, F. & Deems, D. A. A general theory of aversion learning. Ann. N. Y. Acad. Sci. 443, 8–21 (1985).

    Article  CAS  PubMed  Google Scholar 

  16. Alhadeff, A. L. et al. Natural and drug rewards engage distinct pathways that converge on coordinated hypothalamic and reward circuits. Neuron 103, 891–908 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  17. Hernandez, L. & Hoebel, B. G. Feeding and hypothalamic stimulation increase dopamine turnover in the accumbens. Physiol. Behav. 44, 599–606 (1988).

    Article  CAS  PubMed  Google Scholar 

  18. Ferreira, J. G., Tellez, L. A., Ren, X., Yeckel, C. W. & de Araujo, I. E. Regulation of fat intake in the absence of flavour signalling. J. Physiol. 590, 953–972 (2012).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Han, W. et al. A neural circuit for gut-induced reward. Cell 175, 665–678 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Ren, X. et al. Nutrient selection in the absence of taste receptor signaling. J. Neurosci. 30, 8012–8023 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  21. Han, W. et al. Striatal dopamine links gastrointestinal rerouting to altered sweet appetite. Cell Metab. 23, 103–112 (2016).

    Article  PubMed  Google Scholar 

  22. de Araujo, I. E., Schatzker, M. & Small, D. M. Rethinking food reward. Annu. Rev. Psychol. 71, 139–164 (2020).

    Article  PubMed  Google Scholar 

  23. McCutcheon, J. E. The role of dopamine in the pursuit of nutritional value. Physiol. Behav. 152, 408–415 (2015).

    Article  CAS  PubMed  Google Scholar 

  24. de Araujo, I. E. Circuit organization of sugar reinforcement. Physiol. Behav. 164, 473–477 (2016).

    Article  PubMed  PubMed Central  Google Scholar 

  25. Grove, J. C. R. et al. Dopamine subsystems that track internal states. Nature 608, 374–380 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Zhuo, Y. et al. Improved green and red GRAB sensors for monitoring dopaminergic activity in vivo. Nat. Methods 21, 680–691 (2024).

    Article  CAS  PubMed  Google Scholar 

  27. Janak, P. H. & Tye, K. M. From circuits to behaviour in the amygdala. Nature 517, 284–292 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Baxter, M. G. & Murray, E. A. The amygdala and reward. Nat. Rev. Neurosci. 3, 563–573 (2002).

    Article  CAS  PubMed  Google Scholar 

  29. LeDoux, J. E. Emotion circuits in the brain. Annu. Rev. Neurosci. 23, 155–184 (2000).

    Article  CAS  PubMed  Google Scholar 

  30. Maren, S. Neurobiology of Pavlovian fear conditioning. Annu. Rev. Neurosci. 24, 897–931 (2001).

    Article  CAS  PubMed  Google Scholar 

  31. Wassum, K. M. & Izquierdo, A. The basolateral amygdala in reward learning and addiction. Neurosci. Biobehav. Rev. 57, 271–283 (2015).

    Article  PubMed  PubMed Central  Google Scholar 

  32. Namburi, P. et al. A circuit mechanism for differentiating positive and negative associations. Nature 520, 675–678 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  33. Beyeler, A. et al. Divergent routing of positive and negative information from the amygdala during memory retrieval. Neuron 90, 348–361 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  34. Beyeler, A. et al. Organization of valence-encoding and projection-defined neurons in the basolateral amygdala. Cell Rep. 22, 905–918 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Krettek, J. E. & Price, J. L. A description of the amygdaloid complex in the rat and cat with observations on intra-amygdaloid axonal connections. J. Comp. Neurol. 178, 255–279 (1978).

    Article  CAS  PubMed  Google Scholar 

  36. Wang, L. et al. The coding of valence and identity in the mammalian taste system. Nature 558, 127–131 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  37. Hintiryan, H. et al. Connectivity characterization of the mouse basolateral amygdalar complex. Nat. Commun. 12, 2859 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  38. Lutas, A. et al. State-specific gating of salient cues by midbrain dopaminergic input to basal amygdala. Nat. Neurosci. 22, 1820–1833 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Sclafani, A., Koepsell, H. & Ackroff, K. SGLT1 sugar transporter/sensor is required for post-oral glucose appetition. Am. J. Physiol. Regul. Integr. Comp. Physiol. 310, R631–R639 (2016).

    Article  PubMed  PubMed Central  Google Scholar 

  40. Zukerman, S., Ackroff, K. & Sclafani, A. Post-oral appetite stimulation by sugars and nonmetabolizable sugar analogs. Am. J. Physiol. Regul. Integr. Comp. Physiol. 305, R840–R853 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. Ackroff, K. & Sclafani, A. Post-oral fat stimulation of intake and conditioned flavor preference in C57BL/6J mice: a concentration-response study. Physiol. Behav. 129, 64–72 (2014).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Sclafani, A., Zukerman, S. & Ackroff, K. GPR40 and GPR120 fatty acid sensors are critical for postoral but not oral mediation of fat preferences in the mouse. Am. J. Physiol. Regul. Integr. Comp. Physiol. 305, R1490–R1497 (2013).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  43. Sclafani, A., Ackroff, K. & Abumrad, N. A. CD36 gene deletion reduces fat preference and intake but not post-oral fat conditioning in mice. Am. J. Physiol. Regul. Integr. Comp. Physiol. 293, R1823–R1832 (2007).

    Article  CAS  PubMed  Google Scholar 

  44. Fernandes, A. B. et al. Postingestive modulation of food seeking depends on vagus-mediated dopamine neuron activity. Neuron 106, 778–788 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  45. Poulin, J.-F. et al. Mapping projections of molecularly defined dopamine neuron subtypes using intersectional genetic approaches. Nat. Neurosci. 21, 1260–1271 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  46. Leshan, R. L. et al. Ventral tegmental area leptin receptor neurons specifically project to and regulate cocaine- and amphetamine-regulated transcript neurons of the extended central amygdala. J. Neurosci. 30, 5713–5723 (2010).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  47. Flagel, S. B. et al. A selective role for dopamine in stimulus–reward learning. Nature 469, 53–57 (2011).

    Article  CAS  PubMed  Google Scholar 

  48. Saddoris, M. P., Cacciapaglia, F., Wightman, R. M. & Carelli, R. M. Differential dopamine release dynamics in the nucleus accumbens core and shell reveal complementary signals for error prediction and incentive motivation. J. Neurosci. 35, 11572–11582 (2015).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Sias, A. C. et al. Dopamine projections to the basolateral amygdala drive the encoding of identity-specific reward memories. Nat. Neurosci. 27, 728–736 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  50. Han, X. et al. Role of dopamine projections from ventral tegmental area to nucleus accumbens and medial prefrontal cortex in reinforcement behaviors assessed using optogenetic manipulation. Metab. Brain Dis. 32, 1491–1502 (2017).

    Article  CAS  PubMed  Google Scholar 

  51. Boekhoudt, L. et al. Does activation of midbrain dopamine neurons promote or reduce feeding? Int. J. Obes. 41, 1131–1140 (2017).

    Article  CAS  Google Scholar 

  52. Wei, X. et al. Dopamine D1 or D2 receptor-expressing neurons in the central nervous system. Addict. Biol. 23, 569–584 (2018).

    Article  CAS  PubMed  Google Scholar 

  53. Livneh, Y. et al. Estimation of current and future physiological states in insular cortex. Neuron 105, 1094–1111 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  54. Baas, D., Aleman, A. & Kahn, R. S. Lateralization of amygdala activation: a systematic review of functional neuroimaging studies. Brain Res. Rev. 45, 96–103 (2004).

    Article  PubMed  Google Scholar 

  55. Allen, H. N., Bobnar, H. J. & Kolber, B. J. Left and right hemispheric lateralization of the amygdala in pain. Prog. Neurobiol. 196, 101891 (2021).

    Article  PubMed  Google Scholar 

  56. Thanarajah, S. E. et al. Food intake recruits orosensory and post-ingestive dopaminergic circuits to affect eating desire in humans. Cell Metab. 29, 695–706 (2019).

    Article  CAS  PubMed  Google Scholar 

  57. de Araujo, A. M. et al. Asymmetric control of food intake by left and right vagal sensory neurons. Preprint at bioRxiv https://doi.org/10.1101/2023.05.08.539627 (2023).

  58. Qian, K., Liu, J., Cao, Y., Yang, J. & Qiu, S. Intraperitoneal injection of lithium chloride induces lateralized activation of the insular cortex in adult mice. Mol. Brain 14, 71 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  59. Wu, Y. et al. The anterior insular cortex unilaterally controls feeding in response to aversive visceral stimuli in mice. Nat. Commun. 11, 640 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  60. Yagishita, S. et al. A critical time window for dopamine actions on the structural plasticity of dendritic spines. Science 345, 1616–1620 (2014).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  61. Cantu, D. A. et al. EZcalcium: open-source toolbox for analysis of calcium imaging data. Front. Neural Circuits 14, 25 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  62. Pnevmatikakis, E. A. & Giovannucci, A. NoRMCorre: an online algorithm for piecewise rigid motion correction of calcium imaging data. J. Neurosci. Methods 291, 83–94 (2017).

    Article  CAS  PubMed  Google Scholar 

  63. Giovannucci, A. et al. CaImAn an open source tool for scalable calcium imaging data analysis. eLife 8, e38173 (2019).

    Article  PubMed  PubMed Central  Google Scholar 

  64. Boughter, J. D. Jr, Baird, J.-P., Bryant, J., St John, S. J. & Heck, D. C57BL/6J and DBA/2J mice vary in lick rate and ingestive microstructure. Genes Brain Behav. 6, 619–627 (2007).

    Article  PubMed  Google Scholar 

  65. Lever, T. E. et al. Adapting human videofluoroscopic swallow study methods to detect and characterize dysphagia in murine disease models. J. Vis. Exp. 97, e52319 https://doi.org/10.3791/52319 (2015).

  66. McConnell, E. L., Basit, A. W. & Murdan, S. Measurements of rat and mouse gastrointestinal pH, fluid and lymphoid tissue, and implications for in-vivo experiments. J. Pharm. Pharmacol. 60, 63–70 (2008).

    Article  CAS  PubMed  Google Scholar 

  67. Bai, L. et al. Genetic identification of vagal sensory neurons that control feeding. Cell 179, 1129–1143.e23 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  68. Zimmerman, C. A. et al. A gut-to-brain signal of fluid osmolarity controls thirst satiation. Nature https://doi.org/10.1038/s41586-019-1066-x (2019).

    Article  PubMed  PubMed Central  Google Scholar 

  69. Jenkinson, M., Bannister, P., Brady, M. & Smith, S. Improved optimization for the robust and accurate linear registration and motion correction of brain images. Neuroimage 17, 825–841 (2002).

    Article  PubMed  Google Scholar 

  70. Andersson, J. L. R., Skare, S. & Ashburner, J. How to correct susceptibility distortions in spin-echo echo-planar images: application to diffusion tensor imaging. Neuroimage 20, 870–888 (2003).

    Article  PubMed  Google Scholar 

  71. Friston, K. J., Williams, S., Howard, R., Frackowiak, R. S. J. & Turner, R. Movement-related effects in fMRI time-series. Magn. Reson. Med. 35, 346–355 (1996).

    Article  CAS  PubMed  Google Scholar 

  72. Amunts, K., Mohlberg, H., Bludau, S. & Zilles, K. Julich-Brain: a 3D probabilistic atlas of the human brain’s cytoarchitecture. Science 369, 988–992 (2020).

    Article  CAS  PubMed  Google Scholar 

  73. Yates, S. C. et al. QUINT: workflow for quantification and spatial analysis of features in histological images from rodent brain. Front. Neuroinform. 13, 75 https://doi.org/10.3389/fninf.2019.00075 (2019).

  74. Grove, J. C. R. et al. Data for ‘The neural basis of flavour–nutrient learning’. Dryad https://doi.org/10.5061/dryad.h9w0vt4x2 (2026).

Download references

Acknowledgements

We thank K. Yackle and I. Bachmutsky for providing AAV-Con/Fon-ChR2 viruses; and A. MacDonald for his help establishing the vagotomy method.

Funding

J.C.R.G. received funding from the US National Institutes of Health (F31-NS120468); Z.A.K., from the US National Institutes of Health (R01-DK106399, R01-DK138127, R01-NS116626, R01-DK145100) and Howard Hughes Medical Institute (Investigator Award; https://ror.org/006w34k90); and M.T., from Deutsche Forschungsgemeinschaft under Germany’s Excellence Strategy (EXEC 2030–390661388), SFB 1451 (project ID: 431549029, C06) and the German Centre for Diabetes Research (project ID: 82DZD05H1G).

Author information

Authors and Affiliations

  1. Department of Physiology, University of California, San Francisco, San Francisco, CA, USA

    James C. R. Grove, Anna M. Hakimi, Zhengya Liu & Zachary A. Knight

  2. Kavli Institute for Fundamental Neuroscience, University of California, San Francisco, San Francisco, CA, USA

    James C. R. Grove, Anna M. Hakimi, Zhengya Liu & Zachary A. Knight

  3. Neuroscience Graduate Program, University of California, San Francisco, San Francisco, CA, USA

    James C. R. Grove, Anna M. Hakimi, Zhengya Liu & Zachary A. Knight

  4. Howard Hughes Medical Institute, Chevy Chase, MD, USA

    James C. R. Grove, Queenie Li, Jingkun Zhang, Jaewon Choi, Violeta Ubadiah & Zachary A. Knight

  5. Max Planck Institute for Metabolism Research, Cologne, Germany

    Heiko Backes, Bojana Kuzmanovic & Marc Tittgemeyer

  6. Department of Medicine, University of California, San Francisco, San Francisco, CA, USA

    Longhui Qiu

  7. Department of Psychiatry, Yale University School of Medicine, New Haven, CT, USA

    Dana M. Small

  8. Department of Psychology, Yale University School of Medicine, New Haven, CT, USA

    Dana M. Small

  9. University of Cologne, Faculty of Medicine and University Hospital Cologne, Cologne Excellence Cluster on Ageing and Ageing-Associated Diseases (CECAD), Cologne, Germany

    Marc Tittgemeyer

  10. Department of Neurology and Neurosurgery, McConnell Brain Imaging Centre, Montreal Neurological Institute, Modern Diet and Physiology Research Centre, McGill University, Montreal, Quebec, Canada

    Marc Tittgemeyer

  11. Weill Institute for Neurosciences, University of California, San Francisco, San Francisco, CA, USA

    Zachary A. Knight

Authors

  1. James C. R. Grove
  2. Anna M. Hakimi
  3. Queenie Li
  4. Jingkun Zhang
  5. Heiko Backes
  6. Bojana Kuzmanovic
  7. Jaewon Choi
  8. Violeta Ubadiah
  9. Longhui Qiu
  10. Zhengya Liu
  11. Dana M. Small
  12. Marc Tittgemeyer
  13. Zachary A. Knight

Contributions

J.C.R.G. and Z.A.K. conceptualized the study and wrote the manuscript with input from D.M.S. and M.T. All of the experiments performed in mice were designed, analysed and interpreted by J.C.R.G. and Z.A.K. i.g. surgeries were performed by J.C.R.G. and L.Q. All other surgeries were performed by J.C.R.G. Mouse behaviour and in vivo imaging studies were performed by J.C.R.G. and Q.L. Lateralized network tracing studies were performed by J.C.R.G., A.M.H., J.Z. and Z.L. All of the other histological experiments were performed by J.C.R.G., A.M.H., J.C. and V.U. All experiments done in humans were designed, run, analysed and interpreted by H.B., B.K., D.M.S. and M.T.

Corresponding author

Correspondence to Zachary A. Knight.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature thanks Mark Andermann and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Fibre localization for VTA-DA inhibition experiments.

a, Fibre placement in VTA, shown with median slice along anteroposterior (AP) axis (black=littermate controls, red=GtACR). b, Mice with fibres above posterior VTA showed greater inhibition of flavour learning, consistent with the location of the BLA projecting population (Lammel et al.,2008). c, Fibre placement along the mediolateral (ML) axis did not affect learning, as deviations would cancel within animals due to the fixed bilateral spacing of implants. d, Fibre placement along the dorsoventral (DV) axis did not affect learning. ns P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001. Shaded areas show standard error of the mean (SEM). See Supplementary Table 1 for statistical tests, n, and p-values.

Extended Data Fig. 2 Dopamine dynamics across brain regions during IG infusions.

a–e, Two-step normalization procedure. Step 1: fluorescence traces normalized to activity-independent signal (420 nm), removing movement artefacts. Step 2: traces during IG or IP manipulations further normalized to sham infusion or saline injection, correcting for bleaching and long-term drift. Outside this figure, only step 1 normalization was applied to activity traces, and step 2 was applied only to summary plots. a, Example raw fluorescence traces from a single mBLA GRAB-DA site during Ensure IG (top) or sham (bottom). b, Dopamine-dependent signal after 470 nm traces were normalized to 420 nm. c, Further normalization by subtracting sham traces. d, Summary of mBLA dopamine release during Ensure IG vs sham after step 1. e, Summary of the same data after both steps. f, GRAB-DA recordings in mBLA; dotted lines mark external capsule (scale bar, 0.5 mm). g, Mean traces (left) and summary plots (middle, right) showing greater mBLA DA release during and after Ensure infusion. h, No sex differences in mBLA DA release during (left) or after (right) Ensure infusion. i, Glucose, but not mannitol, increases mBLA DA release. j, Transit times for IG infusions: water 12 ± 1 s; Ensure 21 ± 1 s. k, Example traces with sigmoid fits to estimate T5% (orange) and T3σ (blue). l, Summary rise times relative to infusion onset, corrected for transit time (from e): T5% = 54 ± 9 s; T3σ = 41 ± 12 s; mean = 48 ± 11 s. m, Example traces with sigmoid fits to estimate T95% (orange) and T3σ (blue). n, Summary decay times relative to infusion end: T95% = 749 ± 51 s; T3σ = 388 ± 100 s; mean = 569 ± 79 s. o, GRAB-DA recordings in NAc core; dotted lines mark anterior commissure (scale bar, 0.5 mm). p, No change in NAc DA release with Ensure or water. q, GRAB-DA recordings in DLS; dotted lines mark corpus callosum (scale bar, 0.5 mm). r, No change in DLS DA release with Ensure or water. s, Reconstructed fibre placements in BLA (colour scale = ∆F/F during Ensure IG), showing greater DA release in magnocellular vs parvocellular BLA. t, GRAB-DA recordings in lateral BLA; dotted lines mark external capsule (scale bar, 0.5 mm). u, Lateral BLA shows increased DA release during Ensure infusion, but not after. v, Both glucose and mannitol increase DA release in lateral BLA. ns P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001. Shaded areas in mean traces and error bars in bar plots show SEM. See Supplementary Table 2 for statistical tests, n, and p-values.

Extended Data Fig. 3 Dopamine dynamics across brain regions during natural consumption.

a, Fibre placement and viral spread for mBLA recordings (left). Mean GRAB-DA responses during first lick bout of Ensure and water (middle, right). b, mBLA DA peak time is delayed following Ensure vs water. c, No sex differences in mBLA DA release during first Ensure lick bout. d, Summary plots of mBLA DA release during first lick bouts. e, Mean traces (left) and summary plots (middle, right) showing long-term mBLA DA increases during Ensure consumption. f, Fibre placement and viral spread for lateral BLA recordings (left). Mean GRAB-DA responses during first lick bout of Ensure and water (middle, right). g, Peak time of lateral BLA DA release. h, Summary plots of lateral BLA DA release during first lick bouts. i, Mean traces (left) and summary plots (middle, right) showing long-term lateral BLA DA increases during Ensure consumption. j, Fibre placement and viral spread for NAc recordings (left). Mean GRAB-DA responses during first lick bout of Ensure and water (middle, right). k, Peak time of NAc DA release. l, Summary plots of NAc DA release during first lick bouts. m, Mean traces (left) and summary plots (middle, right) showing no long-term NAc DA increase during Ensure consumption. n, Fibre placement and viral spread for DLS recordings (left). Mean GRAB-DA responses during first lick bout of Ensure and water (middle, right). o, Peak time of DLS DA release. p, Summary plots of DLS DA release during first lick bouts. q, Traces (left) and summary plots (middle, right) showing long-term DLS DA increases during Ensure consumption. ns P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001. Shaded areas in mean traces and error bars in bar plots show SEM. See Supplementary Table 2 for statistical tests, n, and p-values.

Extended Data Fig. 4 Dopamine release depends on nutrients in anterior, but not posterior BLA.

a–l, Dopamine (DA) recordings in posterior BLA (pBLA). a, Fibre placement and viral spread for aBLA recordings (left; scale bar, 0.5 mm). DA dynamics during first lick bout of water and Ensure (middle) with summary plot showing greater release to sweet solutions vs water (right). b, Traces (left) and summary plots (right) of DA release during Ensure or water access. c, Summary plots during access to calorie-matched diets: diet A (55%/14%/31% sugar/fat/protein, matches Ensure) and diet B (20%/20%/60%, matches mouse preference). d, Traces (left) and summary plots (right) showing no difference in DA response to Ensure IG in fed vs fasted mice. e, Traces (left) and summary plots (right) of DA release during peptide IG. f, Traces (left) and summary plots (right) of DA release during intralipid vs mineral oil IG. g, Traces (left) and summary plots (right) of DA responses to glucose and sugar analogues during IG. h, Summary plot of DA responses during water or glucose IG with or without SGLT1 antagonist mizagliflozin (Mz). i, Summary plot of DA responses during water or Ensure IG with or without mizagliflozin. j, Summary plot of DA responses during water or intralipid IG with or without antagonists of GPR40 (GW1100, G), GPR120 (AH7614, A), or CD36 (sulfosuc-cinimidyl oleate, S). k, Traces (left) and summary plots (right) of DA release during IP injection of CCK and other hormones. l, Summary plot of DA release during water or intralipid IG with or without prior CCK antagonist devazepide (Dev). m–s, Dopamine (DA) recordings in posterior BLA (pBLA). m, (Left) Fibre placement and viral spread for pBLA recordings (scale bar, 0.5 mm). (Right) DA dynamics during first lick bout of water and Ensure. n, No difference in DA release during bouts of water vs nutrients. o, Traces (left) and summary plots (right) showing increased DA release after IG sugar infusion, independent of SGLT1. p, Traces (left) and summary plots (right) showing no DA increase after nutritive or non-nutritive fat infusions. q, Small but significant DA release during peptide infusion. r, Traces (left) and summary plots (right) showing no DA response to feeding hormones, but significant release to aversive LiCl (IP). s, Increased pBLA DA and decreased aBLA DA following aversive lipopolysaccharide (LPS) injection. ns P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001. Shaded areas in mean traces and error bars in bar plots show SEM. See Supplementary Table 2 for statistical tests, n, and p-values.

Extended Data Fig. 5 DA subpopulation responses across different stages of ingestion.

a–c, Single-cell imaging of all VTA-DA neurons. a, Fibre placement and viral spread for imaging all VTA-DA neurons. b, Dynamics during consumption of Ensure or sham (see Methods). c, Population-weighted activity of activated (magenta) and inhibited (mint) neurons (left) and cumulative distributions (right) showing increased activity during the first Ensure bout (vs sham, p = 2.6e-6; vs water, p = 0.24). d–o, Imaging of VTA-DA/CCK neurons. d, Fibre placement for imaging VTA-DA/CCK neurons (left). Dynamics following water infusion (right). e, Activity rose 36 ± 7 s after Ensure reached the stomach. f, PC1 separates transient vs prolonged subtypes from k-means clustering. g, Population-weighted activity (left) and cumulative distributions (right) showing increased activity after Ensure infusion (vs sham, p = 1.7e-4; vs water, p = 0.038). h, Correlation analyses indicate distinct responses to water vs Ensure infusion. i, Dynamics during consumption of Ensure, water or sham. j, Population-weighted activity (left) and cumulative distributions (right) confirm greater responses during Ensure bouts (vs sham, p = 2.0e-15; vs water, p = 0.0013). k, Binned activity during Ensure bouts. l, Correlation analyses indicate distinct responses to water vs Ensure bouts. m, Overlap of neurons responding to water and Ensure. n, Correlation analyses indicate distinct responses during IG vs natural consumption. o, Overlap of neurons responding during IG and natural consumption. p–z, Imaging of VTA-DA/LepR neurons. p, (Left) Fibre placement for recording VTA-DA/LepR neurons. (Right) Dynamics during water infusion. q, Population-weighted activity (left) and cumulative distributions (right) show activation after water but not Ensure infusion (water vs sham, p = 6.9e-5; water vs Ensure, p = 0.0042; Ensure vs sham, p = 0.33). r, K-means clustering reveals prolonged (purple) and transient (green) subpopulations, shown with mean and representative single-neuron activity. s, Principal component analysis; the first two PCs explain 56% of variance and correspond to activity changes during and after infusion. t, PC1 separates transient vs prolonged subtypes from k-means clustering. u, Correlation analyses indicate distinct responses to water vs Ensure infusion. v, Overlap of neurons responding to water and Ensure infusions. w, Dynamics during consumption of Ensure, water or sham. x, Population-weighted activity (left) and cumulative distributions (right) show increased activity during Ensure bouts (vs sham, 1.2e-6; vs water, p = 0.045). y, Cumulative distributions across subpopulations showed robust activation during Ensure consumption in both DA/CCK (p = 4.2e-11) and DA/LepR (p = 2.3e-12) groups. ns > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Shaded areas show SEM in mean traces and correlation plots. Error bars show SEM in bar plots. See Supplementary Table 3 for statistical tests, n, and p-values.

Extended Data Fig. 6 DA release in neighbouring mBLA and CeA during ingestion.

a–e, Fibre placement and viral spread for recording DA release onto CeA-VGAT neurons (a). DA increased during Ensure IG infusion compared to controls (b) and during the first lick bout (c), with no difference in peak timing across solutions (d). Mean activity remained elevated throughout consumption (e). f–j, Fibre placement and viral spread for recording DA release onto mBLA-Vglut neurons (f). DA increased during Ensure IG infusion (g) and during the first lick bout (h), with peak release delayed for Ensure vs water (i). Mean activity remained elevated throughout consumption (j). ns > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Shaded areas in mean traces and error bars in bar plots show SEM. See Supplementary Table 2 for statistical tests, n, and p-values.

Extended Data Fig. 7 Controls for dopamine projection experiments.

(Top) Viral expression of ChR2 in VTA-DA neurons (or all midbrain DA neurons for DLS terminal experiments) and fibre placement above projection targets (grey=mCherry, red=ChR2; see Fig. 4 for insets). a, Flavour conditioning assay (mCherry controls; Fig. 4b). b, Lick-activated stimulation (mCherry controls; Fig. 4e). c, Lick-terminated stimulation (mCherry controls; Fig. 4f). d, Continuous stimulation (mCherry controls; Fig. 4g). e, Sucrose consumption during CTA training. f, Lick-activated stimulation with water substituted for Ensure. *P < 0.05; **P < 0.01; ***P < 0.001, after Benjamini–Hochberg correction. Error bars show SEM. See Supplementary Table 1 for statistical tests, n, and p-values.

Extended Data Fig. 8 aBLA neurons track sweet taste and nutrient information to drive learning.

a, Protocol for delayed flavour–nutrient conditioning. Mice received 10 min access to a flavoured saccharin solution followed by IG infusion of water or glucose. Mice did not learn a preference if glucose infusion occurred only after licking ceased. b, Fibre placement and viral spread for aBLA recordings. Averaged aBLA neuron dynamics during sucrose bouts (z-score) and distributions of activity compared to sham (p = 6.7 × 10-15, KS test) or mock lick bouts (p = 0.005, KS test). c, Peak (magenta) and 5% rise times (mint) with example cells (left) and summary plots (right). d, Lick count across trials and during the first 0.5 s of bouts. e, Activity of all cells (left) and sucrose-activated cells (right) across trials and during the first 0.5 s of bouts. f, Correlation of activity and lick count during first 0.5 s of sucrose vs mock bouts (sucrose: R² = 0.030, p = 0.0017; mock: R² = 0.00014, p = 0.95). g, Cross-registered neuron dynamics during sucrose (left) and sucralose (right) bouts. h, Mean traces (left) and activity (right) of sucrose-activated neurons during sucrose vs sucralose and sucrose vs water bouts. Minor differences in sucrose activity reflect cross-registration. i, Correlation of activity and lick count during mock, sucrose, sucralose, and water bouts (mock: R² = 0.062, p = 0.21; sucrose: R² = 0.018, p = 0.016; sucralose: R² = 0.17, p = 1 × 10-7; water: R² = 0.076, p = 0.0089). j, PCA of cross-registered neurons across sucrose, sucralose, and water sessions: scree plot (left) and first three PCs during each bout (right). k, Single-neuron dynamics (left) and distribution of activity (right) during and after sucrose IG infusion (both: p = 0.002, KS test). l, Population-weighted z-score activity during IG infusions, averaged by mouse. m, Nutrient specific activation during IG infusion of sucrose, but not water or mannitol. n, Activity after IG infusions. o, Unlike early activation, post-IG activity is not nutrient-selective. p, Population-weighted z-score activity after IG infusions, averaged by mouse. q, Distribution of neurons activated (red), inhibited (blue), or unchanged (black) during early (bouts 2–6) vs late (last five) sucrose bouts. Percent activated given in red. r, Late sucrose bouts (but not sucralose or water) activate these neurons. s, Trial-based drinking sessions extended to 120 trials in a new cohort, which showed nutrient-specific bout responses. ns P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001. Shaded areas in mean traces and error bars in bar plots show SEM. See Supplementary Table 3 for statistical tests, n, and p-values.

Extended Data Fig. 9 aBLA neurons are necessary for flavour-nutrient learning.

a, Preference differences were not explained by fibre localization across the three axes. b, aBLA-D1R inhibition reduced CS+ intake on the first training day. c, Continuous inhibition did not alter consumption of the familiar nutritive solution Ensure. d, Conditioned taste aversion assay; inhibition during training did not block LiCl-induced aversion learning. ns P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001. Shaded areas in correlation plots and error bars in bar plots show SEM. See Supplementary Table 1 for statistical tests, n, and p-values.

Extended Data Fig. 10 Internal nutrients preferentially engage the left BLA.

a–b, Unilateral recordings in left or right BLA. a, Mean DA release during Ensure IG infusion is greater in left BLA (left, mean traces; right, summary plots). b, DA release during the first lick bout is greater in right BLA. c–d, Simultaneous recordings in left and right BLA. c. Example from one mouse showing DA activity in left and right BLA during Ensure IG infusion. d, Example from one mouse at baseline. e, In humans, PET (11C-raclopride) shows significant DA release after milkshake in left BLA but not right BLA (left-right difference not significant). f, fMRI shows significantly stronger activation after glucose consumption in the left relative to the right BLA. g, Left dominance may arise from (1) asymmetric vagal input, or (2) lateralized feedback circuitry. h, Bilateral mBLA dopamine recordings during Ensure IG after unilateral vagotomy (VGX). i, Difference in DA release between left and right mBLA before and after VGX. j, Polysynaptic retrograde tracing mapped inputs to left vs right VTA → BLA dopamine pathways. k, Greater labelling of insular cortex following polysynaptic tracing from left pathway. l, Top fifteen regions showing greater labelling after left vs right BLA injection (normalized to VTA labelling; n = 8 left injections, 6 right injections). m, Force-directed graph illustrating strong connectivity among regions in (l). Mice were sated for all infusion recordings (here and elsewhere), unless otherwise stated. Brains shown in neurological orientation (left = subject’s left). Full list of abbreviations in Extended Data Fig. 12. ns P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001. Shaded areas in mean traces and error bars in bar plots show SEM. See Supplementary Tables 2 and 4 for statistical tests, n, and p-values. See Supplementary Table 5 for raw data from pseudorabies experiment. See Supplementary Table 6 for connectivity matrix.

Extended Data Fig. 11 Nutrient-triggered dopamine release in the right BLA.

a–f, Unilateral right BLA recordings. a, Fibre placement and viral spread for GRAB-DA recording in right BLA (scale bar, 1 mm). b, No sex differences in right BLA DA release during (left) or after (right) Ensure infusion. c, Mean traces (left) and summary plots showing glucose responses in right BLA are nutrient-specific (middle) but smaller than in left BLA (right). d, (Left) Mean trace showing DA release in right BLA during first bout of Ensure or water consumption. (Right) Summary plots of right BLA DA release during first lick bouts. e, No sex differences in right BLA DA release during first Ensure lick bout. f, Traces (left) and summary plots (right) of right BLA DA release during Ensure or water consumption. g–k, Bilateral BLA recordings. g, Fibre placement and viral spread for simultaneous GRAB-DA recording in left and right BLA. h, Spontaneous activity is coordinated (left) and comparable (right) across hemispheres. i, Ensure responses are larger in left BLA. j, Glucose responses are larger in left BLA. k, Intralipid responses are larger in left BLA. l–n, PET and fMRI recordings in humans. l, Number of brain-wide voxels (PET) showing increased dopamine release after milkshake consumption (normalized to water), with two peaks: an immediate oral signal (0–5 min after drinking onset) and a delayed post-ingestive signal (5–10 min after drinking offset). m, Time course of dopamine release in left and right BLA. Significance shown for oral and post-ingestive time bins (grey bars). n, Left and right BLA masks (left) and fMRI activity following glucose consumption (right). Activity was significant in left, but not right BLA, and direct comparison showed a significant difference between left and right BLA (F(1,703) = 7.83, *p* = .005). Green overlay refers to anatomically defined BLA masks by Amunts et al., 2020. o–z, Vagotomy experiments. o, Right vagotomy (rVGX). p, Ensure responses in both hemispheres partially depend on right vagus. q, Glucose responses in both hemispheres partially depend on right vagus. r, Right vagotomy does not affect lateralized glucose responses. s, Left BLA intralipid responses show greater sensitivity to right vagotomy. t, Right vagotomy abolishes lateralized intralipid responses. u, Left vagotomy (lVGX). v, Ensure responses in both hemispheres partially depend on left vagus. w, Left vagotomy does not affect glucose responses in either hemisphere. x, Left vagotomy does not affect lateralized glucose responses. y, Intralipid responses in both hemispheres partially depend on left vagus. z, Left vagotomy does not affect lateralized intralipid responses. ns P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001 (FWE-corrected for human data). Shaded areas in mean traces and error bars in bar plots show SEM. In box plots, top and bottom edges define 25th and 75th percentile; whiskers show maximum and minimum values. See Supplementary Tables 2 and 4 for statistical tests, n, and p-values.

Extended Data Fig. 12 Networks underlying dopamine lateralization.

a, Rabies tracing of DAT+ midbrain neurons projecting to left aBLA. Cre-dependent TVA was expressed in VTA/SNc, followed 3 weeks later by injection of CVS-N2c-dG-mGFP in left aBLA. b, Labelled midbrain cells 1 week after rabies injection, showing strong ipsilateral bias. Insets show a single contralateral VTA cell; some labelling in ipsilateral SNc. c–e, Quantification of labelled cells: 78 ± 7% ipsilateral VTA, 17 ± 8% ipsilateral SNc, 5 ± 2% contralateral VTA, and none in contralateral SNc. f, Cells labelled four days after PRV-Introvert-GFP injection into left BLA and right BLA of DAT-cre mice. No substantial differences observed in the nucleus of the solitary tract (NTS) or arcuate nucleus (ARC), where nutrient signals enter the brain. g, Greater labelling of several amygdalar regions after left-side injection. h, Greater labelling of BNST following right vs left injection (arrows indicate VISC showing greater labelling after left injection). i, Top fifteen regions that demonstrated strongest labelling after injection into right vs left BLA. j, Force-directed graph showing connectivity between areas in (e). k–l, Connectivity matrixes and hierarchical clustering of areas with greatest labelling after injection into left vs right BLA (f) or right vs left BLA (g). Abbreviations (used here and Extended Data Fig. 10): ACVII, accessory facial motor nucleus; AI, agranular insula; ARC, arcuate nucleus; BLA, basolateral amygdala; BMA, basomedial amygdala; BNST, bed nuclei of the stria terminalis; CM, central medial nucleus of the thalamus; COA, cortical amygdala area; DECLIVE, lobule VI (declive); DMH, dorsomedial nucleus of the hypothalamus; ECT, ectorhinal area; GP, globus pallidus; IAM, interanteromedial nucleus of the thalamus; IAD, interanterodorsal nucleus of the thalamus; LA, lateral amygdala; LGv, ventral lateral geniculate; Lob III, lobule III; Lob IX, uvula; MeA, medial amygdala; nBIC, nucleus of the brachium of the inferior colliculus; NTS, nucleus of the solitary tract; PA, posterior amygdala; PAG, periaqueductal grey; PBN, parabrachial nucleus; PERI, perirhinal area; PoL, posterior limiting nucleus of the thalamus; RTN, reticular nucleus of the thalamus; SMT, submedial nucleus of the thalamus; TT, taenia tecta; VAL, ventral anterior-lateral complex of the thalamus; VISC, visceral area; VPL, ventral posterolateral nucleus of the thalamus. ns P > 0.05; *P < 0.05; **P < 0.01; ***P < 0.001. Error bars in bar plots show SEM. See Supplementary Table 5 for raw data from pseudorabies experiment. See Supplementary Table 6 for connectivity matrix.

Supplementary information

Supplementary Information (download PDF )

Legends and additional references for Supplementary Tables 1–6.

Reporting Summary (download PDF )

Supplementary Table 1 (download XLSX )

Behaviour statistics. Summary information for all behavioural tests for each manipulation. Values are reported to 1 decimal place. Number of animals is reported as n. P values are reported for all statistical tests, and are shown in bold if below standard significance level (α = 0.05). In cases in which a Benjamini–Hochberg correction was made for multiple comparisons, an adjusted P value (p′) is shown (where p′ = P × m/i, where i is rank and m is number of cohorts tested; Methods). X′ corresponds to the normalized value of X. For self-stimulation, ‘presses’ means the number of inactive lever presses has been subtracted from the number of active lever presses. In flavour conditioning experiments in which mice began the test with >90% preference for one flavour, statistics are also provided (in parentheses) with these mice excluded. Permutation (perm.) tests use an exact test (or 100,000 iterations if n > 9) and are unpaired, unless otherwise mentioned (note: this means the lowest calculable P value is 1 × 10−5).

Supplementary Table 2 (download XLSX )

Dopamine imaging statistics. Summary information for all recordings of dopamine release in each brain region. Values are reported to 2 decimal places. Number of animals is reported as n. P values are reported for all statistical tests, and are shown in bold if below standard significance level (α = 0.05). X′ corresponds to the normalized value of X. For photometry traces, dF′/F′ means the dF/F values seen during sham infusion have been subtracted. For rise times, s′ means that the time is relative to approximate entry of liquid into the stomach. For fall times, s′ means that the time is relative to the end of the infusion. For peak times during licking, s′ means that the time is relative to the start of the first licking bout. The baseline is defined as the first 10 min of the recording for GI and post-GI measurements, or 0–20 s before the start of the first licking bout for oral measurements. Permutation tests use an exact test (or 100,000 iterations if n >9) and are unpaired, unless otherwise mentioned (note that this means the lowest calculable P value is 1 × 10−5).

Supplementary Table 3 (download XLSX )

Single-cell imaging statistics. Summary information for all 1- and 2-photon miniscope recordings. Values are reported to 2 decimal places. The number (n) of animals and number of cells are reported separately. P values are reported for all statistical tests, and are shown in bold if below standard significance level (α = 0.05). X′ corresponds to the normalized value of X. For rise times, s′ means that the time is relative to either the onset of the licking bout or to the approximate entry of liquid into the stomach. Population-weighted activity (pop-z) is calculated for a group of neurons (for example, activated or inhibited neurons) by multiplying activity by the fraction of neurons belonging to that group. Linear-mixed models (LME) treat mouse differences as both a fixed random effect and covariate (Methods). The relative contribution of mouse differences to total variance is reported by ICCs. Note that, for PCA, the number of PCs is equal to the number of cells. Permutation tests use an exact test (or 100,000 iterations if n > 9) and are unpaired, unless otherwise mentioned (note that this means that the lowest calculable P value for those tests is 1 × 10−5).

Supplementary Table 4 (download XLSX )

fMRI imaging statistics (n = 38). Activity from left and right BLA clusters binned every 2 min (20 min total), as shown in Extended Data Figs. 10f and 11n. For PET statistics, see ref. 56. Each of the ten time bins after glucose intake was contrasted against the baseline (2 min before glucose intake), representing the change in brain activity after glucose intake relative to the baseline. The contrast for the first effect was [ones(1,10)], and for the following contrasts [1], corresponding to the different time bins. The significance threshold was P < 0.05, FWE-corrected for multiple comparisons at the voxel level within the left or right BLA mask, respectively, with initial peak-level threshold set to P < 0.05 uncorrected.

Supplementary Table 5 (download XLSX )

Pseudorabies statistics. Regional labelling (load) was normalized to VTA load to control for injection efficiency. From the 1,328 annotated CCFv3 regions, only grey-matter areas were retained (372 total), with cortical layers collapsed into single structures. Regions smaller than 100 × 100 × 100 pixels were excluded to remove small nuclei with high variance from few labelled cells, yielding 131 analysed regions.

Supplementary Table 6 (download XLSX )

Brain connectivity tables. Connectivity matrices were generated manually, with each pair given a number indicating bidirectional projection strength from 0 (no known direct connection) to 5 (strongest projection known from this area). The connectivity strength between an area and itself was assigned a value of 5. For deciding what strength to assign a projection, large databases like the Allan Mouse Brain Connectivity Atlas (ref. 1 of the Supplementary References) were preferred, supplemented by additional sources comprising reviews and anatomical tracing studies (refs. 2–32 of the Supplementary References).

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Grove, J.C.R., Hakimi, A.M., Li, Q. et al. The neural basis of flavour–nutrient learning. Nature (2026). https://doi.org/10.1038/s41586-026-11229-5

Download citation

  • Received:

  • Accepted:

  • Published:

  • Version of record:

  • DOI: https://doi.org/10.1038/s41586-026-11229-5