A. Personal Statement

Learning new skills requires trying new things, evaluating the outcomes, and modifying future performance. My lab is interested in how these processes are implemented in the brain. To distinguish general principles from behavior-, effector-, and species-specific solutions to motor control problems, we study vocal control in both songbirds and parrots, two species with distinct, independently evolved vocal circuits and learning capacities. We also study motor control by comparing neural mechanisms by which mice aim their limbs and tongues. For all research programs, we combine high channel count awake-behaving electrophysiology, closed-loop optogenetics, and machine learning-guided behavioral analysis. Our guiding philosophy is that comparative approaches in systems neuroscience, though rare, are important for discovering the core functions of ancient, distributed circuits linking cortex, basal ganglia, brainstem and cerebellum.

1. Gadagkar V, Puzerey P, Chen R, Baird-Daniel E, Farhang A, Goldberg JH. Dopamine neurons encode performance error in singing birds. Science. 2016 Dec 9;354(6317):1278-1282.

2. Chen R, Puzerey P, Roeser A, Riccelli T, Podury A, Maher K, Farhang A, Goldberg JH. Songbird ventral pallidum sends diverse performance error signals to dopaminergic midbrain. Neuron. 2019 May 11;doi: 10.1016/j.neuron.2019.04.038.

3. Goldberg JH and Fee MS. A cortical motor nucleus drives the basal ganglia-recipient thalamus in singing birds. Nature Neuroscience. 2012 Feb 12;15(4):620-7.

4. Bollu T, Whitehead SC, Prasad N, Walker J, Shyamkumar N, Subramaniam R, Kardon B, Cohen I, Goldberg JH. Automated home cage training of mice in a hold-still center-out reach task. J Neurophysiol. 2019 Feb 1;121(2):500-512.

5. Bollu T, Whitehead SC, Kardon B, Redd J, Liu MH, Goldberg JH. Tongue Kinematics. Cortex-dependent corrections as the mouse tongue reaches for, and misses, targets. bioRxiv 655852; doi: https://doi.org/10.1101/655852

B. Positions and Honors

6/2019-present Associate Professor, Department of Neurobiology and Behavior, Cornell University

2012-2019 Assistant Professor, Department of Neurobiology and Behavior, Cornell University

2005-2012 Postdoc, Laboratory of Michale Fee, McGovern Institute for Brain Research, M.I.T.

2001 Methods in Computational Neuroscience, Marine Biological Lab, Woods Hole, MA.

1997-2005 Medical Scientist Training Program, Laboratory of Rafael Yuste, Columbia University

1995-1997 Undergraduate research fellow, Laboratory of Daniel Alkon, NINDS, NIH.

Other Experience and Professional Memberships

2017-ongoing, Scientific Advisory Committee, Dystonia Medical Research Foundation

2014-present, Ad hoc member of NSF/NIH/NINDS Study Sections

2014-2016 Grant reviewer, Israel Science Foundation

2017, 2018 Grant reviewer, Human Frontiers Science Program

2015 Co-organizer, Thalamus and corticothalamic interactions, HHMI, Ashburn, VA

2013-16, Reviewer, Computational and Systems Neuroscience Meeting, Salt Lake City, UT

2013, Grant reviewer, Neurological Foundation of New Zealand

2010-ongoing, Member, International basal ganglia society

1999-ongoing, Member, Society for Neuroscience

2012-ongoing, Manuscript Reviewer: Science, Nature, Neuron, Nature Neuroscience, Journal of Neurophysiology, Journal of Neuroscience, Nature Communications, Journal of Physiology, eLife, Royal Proceedings B, Brain and Behavioral Sciences, PLOS Biology, Scientific Reports

Honors

2016 Kavli Fellow

2015-2020 NIH New Innovator

2014-2018 Pew Biomedical Scholar

2013-2016 Klingenstein Fellowship in the Neurosciences

2010-2015 K99/R00 Pathway to Independence Award, NINDS

2009-2010 Charles King Trust Postdoctoral Fellowship

2005-2008 Damon Runyon Research Foundation Postdoctoral Fellowship

2004 Dean’s Award for best Ph.D. thesis, Columbia University

2004 Ph.D. thesis awarded with distinction, Columbia University

1997-2005 Medical Scientist Training Program Scholarship

1997 Phi Beta Kappa Society, Haverford College

1997 Honors Thesis in Biology, Haverford College

C. Contributions to Science

1. Discovery of performance error signals in singing birds. An unanswered question was how performance is evaluated as ‘good’ or ‘bad’ during practice. It was well established that dopamine activity contributes to reinforcement learning by encoding reward prediction error in tasks where animals learn for primary rewards such as food or juice. But it was unknown if dopamine could also encode performance error in tasks that did not involve reward. To test if dopamine encodes performance error, we recorded dopamine neurons in singing birds while controlling perceived error with distorted auditory feedback. Remarkably, dopamine activity encoded performance just like reward: phasic bursts following better-than-predicted outcomes, and phasic pauses following worse-than-predicted ones. We next identified origins of dopaminergic error signals in a ventral pallidal region outside the classic song system, revealing a previously unidentified ‘actor/critic’ circuit motif inside the songbird brain. This discovery was high impact because it demonstrated that circuit motif associated with drug addiction and foraging in mammals is ancestral and can be repurposed for learning a motor sequence like birdsong. Finally, dopamine is, of course, not the only neuromodulator likely to be important for a complex behavior like birdsong learning. The basal forebrain also has strong cholinergic projections to motor cortical areas, yet the role of acetylcholine for motor learning remained unclear. We found that manipulation of cholinergic signaling in vocal motor cortex of juvenile birds did not affect vocal babbling, yet chronic blockade over weeks impaired learning, resulting in an impoverished song with excess variability, abnormal acoustic features and reduced similarity to tutor song.

a. Fee MS and Goldberg JH. A hypothesis for basal ganglia dependent reinforcement learning in the songbird. Neuroscience. 2011 Dec 15;198:152-70.

b. Gadagkar V, Puzerey P, Chen R, Baird-Daniel E, Farhang A, Goldberg JH. Dopamine neurons encode performance error in singing birds. Science. 2016 Dec 9;354(6317):1278-1282.

c. Puzerey P, Maher K, Chen R, Prasad N, Goldberg JH. Vocal learning in songbirds requires cholinergic signaling in a motor cortex-like nucleus. J Neurophysiol. 2018 Oct 1;120(4):1796-1806.

d. Chen R, Puzerey PA, Roeser AC, Riccelli TE, Podury A, Maher K, Farhang AR, Goldberg JH. Songbird Ventral Pallidum Sends Diverse Performance Error Signals to Dopaminergic Midbrain. Neuron. 2019 May 11;doi: 10.1016/j.neuron.2019.04.038.

e. Murdoch D, Chen R, Goldberg JH. Place preference and vocal learning rely on distinct reinforcers in songbirds. Scientific Reports. 2018 Apr 30;8(1):6766.

2. High-resolution tracking of mouse tongue and limb. We developed a suites of technologies that combine high resolution sensing of mouse forelimb, automated high-throughput training in the homecage, and closed-loop optogenetic manipulation of neural activity in freely moving mice. We trained mice in a complex center-out reach task, inspired by past work in primates, that required mice to carve complex reach sequences, or trajectories through space. We found that motor cortex inactivation did not affect sequence timing, direction, or trajectory shape, but uniformly reduced the peak speed of kinematic primitives - the units of movement in a sequence. Thus, trajectories during motor cortical inactivation were ‘shrunk,’ as if the letter ‘N’ was drawn as a miniature ‘N’. To test if this result would generalize to a different motor effector, we combined kilohertz frame-rate imaging and novel deep learning based artificial neural network, to tracke the rodent tongue in 3D at decamicron-millisecond spatiotemporal precision. Cue-evoked licks exhibited previously unobserved fine-scale movements which, like a primate hand searching for an unseen object, were produced after tongue protrusions and were directionally biased towards remembered spout locations. Tongue motor cortex inactivation abolished these fine-scale adjustments, resulting in well-aimed but hypometric licks that missed the spout. Together, our results showed cortical inactivations cuased hypometria of both limb and tongue. By watching the tongue in action for the first time, we additionally discovered that licks cannot be explained by open loop central pattern generators that drive simple binary ballistic events. Instead, individual licks exhibited complex, variable trajectories with limb-like dynamics, including the production of motor cortex-dependent online adjustments that facilitate target contact.

a. Bollu T, Whitehead SC, Prasad N, Walker J, Shyamkumar N, Subramaniam R, Kardon B, Cohen I, Goldberg JH. Automated home cage training of mice in a hold-still center-out reach task. J Neurophysiol. 2019 Feb 1;121(2):500-512.

b. Bollu T, Whitehead SC, Kardon B, Redd J, Liu MH, Goldberg JH. Tongue Kinematics. Cortex-dependent corrections as the mouse tongue reaches for, and misses, targets. bioRxiv 655852; doi: https://doi.org/10.1101/655852

c. Bollu T, Whitehead SC, Prasad N, Walker J, Shyamkumar N, Subramaniam R, Kardon B, Cohen I, Goldberg JH. Motor cortical inactivation reduces the gain of kinematic primitives in mice performing a hold-still center-out reach task. bioRxiv 304907; doi: https://doi.org/10.1101/304907

3. Determining how signals propagate through songbird BG-thalamic circuits during behavior. In past work and in the first paper from my own lab, I performed the first recordings from all six BG classes in singing juvenile birds, resulting in a complete classification of intrinsic BG circuitry, identification of sparse temporally precise discharge in striatal medium spiny neurons, the discovery of two pallidal pathways homologous to primates, and a new model of BG output to the thalamus. The BG have several cell classes, but one main output: GABAergic pallidal neurons that project to the thalamus. In the classical model, thalamic activity is suppressed by the tonic inhibition of these pallidal inputs, and pauses in pallidal firing enable thalamic activation. My studies overturned these assumptions. I recorded, for the first time, simultaneously from connected pairs of pallidal terminals and thalamic neurons during behavior, and found that (1) Connected pallidal and thalamic neurons fired in concert at high rates (>200 Hz). (2) Thalamic spiking was not restricted to pallidal pauses. (3) Thalamic spikes were time-locked to pallidal spikes with submillisecond precision. Together, these findings revised the predominant model that BG outputs silence thalamic activity, and instead showed that the BG control thalamic spike timing.

d. Goldberg JH, Farries MA and Fee MS. Integration of cortical and pallidal inputs in the basal ganglia-recipient thalamus of singing birds. Journal of Neurophysiology 2012 Sep;108(5):1403-29.

e. Goldberg JH, Farries MA and Fee MS. Basal ganglia output to the thalamus: still a paradox. Trends Neurosci. 2013 Dec;36(12):695-705.

f. Goldberg JH, Adler A, Bergman H, and Fee MS. Singing related neural activity distinguishes two classes of putative pallidal neuron in the songbird basal ganglia: Comparison to the primate internal and external pallidal segments. Journal of Neuroscience. 2010 May 19;30(20):7088-98

g. Goldberg JH and Fee MS. Singing related neural activity distinguishes four classes of putative striatal neuron in the songbird basal ganglia. Journal of Neurophysiology. 2010 Apr;103(4):2002-14.

h. Pidoux M, Bollu T, Riccelli T, Goldberg JH. Origins of basal ganglia output signals in singing juvenile birds. J Neurophysiol. 2014 Nov 12:jn.00635.2014. doi: 10.1152/jn.00635.2014.

4. Determining neural circuits driving motor variability during learning. The basal ganglia (BG), thalamus and cortex form an evolutionarily conserved brain loop widely implicated in human disease and trial-and-error learning. The first big question I addressed in songbirds, a tractable model system for BG research, relates to how exploratory behaviors are generated – the ‘trial’ part of learning. Juvenile songbirds babble, and the variability of their vocalizations is thought to represent motor exploration essential for learning. It was long assumed that the uncoordinated, variable motor behavior of juveniles was the de facto output of an immature motor system. Yet I found that variability is instead actively generated by BG-related circuits. Specifically, lesions to the motor thalamus abolished vocal variability, resulting in a highly stereotyped song. It was as if a beginner tennis player, spraying balls all over the court, was suddenly hitting the same shot over and over. Yet lesioning the BG inputs to the motor thalamus did not affect babbling at all. This result was surprising because canonical models assumed that the thalamus was principally a relay from BG to cortex, and yet I observed that thalamic motor function did not require the BG. I next performed the first simultaneous recordings of thalamic neurons and their BG inputs in any model system, allowing me to analyze the origins of thalamic premotor signals. I discovered that inputs to the thalamus from a motor cortical nucleus, and not inputs from the BG, drove thalamic premotor activity during babbling. Activity in the homologous part of the human thalamus controls movement initiation and drives Parkinsonian tremor, and was thought to be primarily controlled by BG inputs. Yet these studies suggest a different view: motor cortex, not the BG, drives thalamic premotor signals during behavior, and a previously overlooked cortico-thalamocortical loop generates motor exploration during learning.

a. Goldberg JH and Fee MS. Vocal babbling in songbirds requires the basal ganglia-recipient motor thalamus but not the basal ganglia. Journal of Neurophysiology. 2011 Jun;105(6):2729-39.

b. Aronov D, Veit L, Goldberg JH, Fee MS. Two distinct modes of forebrain circuit dynamics underlie temporal patterning in the vocalizations of young songbirds. Journal of Neuroscience. 2011 Nov 9;31(45): 16353-68.

c. Olveczky BP, Otchy T, Goldberg JH, Aronov D, Fee MS. Changes in the neural control of a complex motor sequence during learning. Journal of Neurophysiology. 2011 Jul;106(1):3867.

d. Goldberg JH and Fee MS. A cortical motor nucleus drives the basal ganglia-recipient thalamus in singing birds. Nature Neuroscience. 2012 Feb 12;15(4):620-7.

5. Identifying ‘canonical’ microcircuits and associated dendritic functions in the neocortex. My PhD dissertation in Rafael Yuste’s lab at Columbia University focused on inhibitory microcircuits of the cortex. I tested the big idea that the cortex has may be composed of one canonical microcircuit repeated millions of times over the surface of the brain, i.e. it has a modular organization. If this is the case, then studying any one module could reveal basic principles about the entire structure. To test this hypothesis, I used electrophysiology, two photon calcium imaging, immunohistochemistry and cellular reconstructions, to test whether microcircuitry varied across widespread cortical regions. For example, I discovered that specialized potassium channels restrict dendritic calcium accumulations in one cell class, while low voltage activated calcium channels mediate global calcium signals in others. Meanwhile, Ca-permeable AMPA receptors contribute to calcium microdomains in aspiny dendrites of PV interneurons across cortex. This body of work showed, for the first time, that the dendrites of inhibitory interneurons were active structures, expressing a wide variety of voltage gated and synaptic ion channels in a cell class specific manner. Further, interneuronal subclass specializations were invariant across visual, sensory and motor cortices, supporting the overarching hypothesis of modular cortical architecture. This modular view of brain function influenced my thinking in the clinic. When I confronted the complexity of neurological disease in medical school, I decided that the ideal way to understand disease mechanims would be to isolate one basal ganglia module and study it in relation to the specific behavior it controls. This is why I chose the songbird: it is a model system with a discrete basal ganglia module devoted to natural trial and error motor learning.

a. Goldberg JH, Yuste R, Tamas G. Ca2+ imaging of mouse neocortical interneuron dendrites: Contribution of Ca2+-permeable AMPA and NMDA receptors to subthreshold Ca2+dynamics. Journal of Physiology (Previewed article). 2003 Aug 15;551:67-78.

b. Goldberg JH, Tamas G, Yuste R. Ca2+ imaging of mouse neocortical interneuron dendrites: Ia-type K+ channels control action potential backpropagation. Journal of Physiology (Previewed article). 2003 Aug 15;551:49-65.

c. Goldberg JH, Lacefield CO, Yuste R. Global dendritic calcium spikes in mouse layer 5 low-threshold spiking (LTS) interneurones: implications for control of pyramidal bursting. Journal of Physiology. 2004 Jul 15;558(Pt 2):465-78.

d. Goldberg JH and Yuste R. Space Matters: Local and global dendritic calcium comparmentalization in cortical interneurons. Trends in Neurosciences. 2005 28: 158-167.

e. Goldberg JH, Tamas G, Aronov D, Yuste R. Calcium microdomains in aspiny dendrites. Neuron (Cover article). 2003 Nov 13;40(4):807-21.