The key element of decision making is the commitment to an action: all-or-none, both sustaining a decision and locking out competing action. In contrast, the choice part of decision-making is less important. Even a simple random choice or first choice is an effective decision mechanism, but losing out the all-or-none means the decision isn’t a decision. The key is ensuring that once a choice is made, the animal sticks to that choice. The losing action should not interfere with the winner. Specifically a decision suppresses dithering: switching between competition actions [Redgrave et al 1999].
This essay uses wall-following from essay 41 as the decision. If the animal detects a wall to the right, it will follow that wall for a time, improving search over random walk by reducing the search to a single dimension. In this case, the choice is left or right, which particularly matters because the choice requires communication between the two sides, which requires specific circuits because commissures are relatively rare.
Decision: two-phase commit
Decision-making has two main components: the choice (preparation) and commitment. A decision that doesn’t sustain or that doesn’t lock out competing stimuli isn’t a decision. Decision-making can split into a preparation / selection phase, which compare options — taking time if necessary, followed by a commit winner-take-all phase where the winning action goes forward and any losing action is locked out.
Most decision research is focused on the preparation phase because people are interested in choosing A vs B, and much less research on the commitment implementation, the timeout and lock-out. For the essay simulation, the commitment is more important and needs to be implemented first. Without the commitment, a losing option can continually interrupt the animal, distracting it from its goals. The requirements for commitment are something like:
- Sustain
- Timeout: prevent the sustain from becoming perseveration
- Lockout: prevent competing actions
Orientation and wall-following
For a decision, this essay uses wall-following (thigmotaxis), continuing from essay 41. Wall-following needs to be treated as a decision beyond a single swimming cycle. Consider the alternative where a lateral-line sense is a simple sensory-action reflex for each swimming cycle. Without a longer conception of the decision, the animal can’t avoid perseveration: it would circle a pillar or a convex arena endlessly. Any timeout needs to curtail wall-following, not right turns. Similarly, without persistence the animal might alternate left and right wall-following in a crowded environment, where both the left and right lateral-line indicates obstacles. Of the two, the timeout issue is more critical, but the ability to continue an action, is necessary to enable a “win-stay” strategy.
Because the research hasn’t located the circuit for wall following, these essays need to choose a brain location for it. The previous essay proposed R1.a (anterior hindbrain) as the driver of thingmotaxis, but an alternative uses OT (optic tectum) as an orientation center for thigmotaxis. Because OT receives lateral-line input via M.ts (torus semicircularis / inferior colliculus) [Zeymer et al 2018], it has the sensory information needed to turn toward a wall. OT is known as an orientation center. When a surprising or salient sensation appears, the animal turns toward it. If we imagine the proto-vertebrate as non-cortical, then that orient is likely OT. Even in mammals with a strongly developed cortex that can provide orientation functionality, OT is perhaps the strongest [Schall 2019].
One immediate question is if the orientation also implements sustain. Compare the left and right model. In the left model, the orientation system implements sustain itself, driving and sustaining a turn action. The right model has a distinct sustain system, which may be driven by motor efference copies. The known connectivity of OT could support either model.
OT has independent sustaining capabilities, in part due to sodium channel modulation [Ghitani et al 2016], [Thompson AC and Aizenman 2023]. OT also has a loop with T.pf (parafascicular thalamus) and S.d (dorsal striatum), which can implement the timeout using A2a.s (adenosine G-s coupled stimulatory receptor) and adenosine accumulation. Although the left model is possible, several studies report OT burst neurons activate at the decision point [Lintz et al 2019], [Stine et al 2023], with ramping neurons before the decision and action [Munoz and Wurtz 1995], [Lintz et al 2019], not after the decision, suggesting the model on the right. Ppt (pedunculopontine tegmental nucleus) is well-suited as a central node of the sustain role. Ppt maintains activity from one decision to another in tasks that repeat decisions [Thompson JA et al 2016]. It also receives widespread input from hindbrain motor areas, including R1.a and R5.my.gi (medulla giganocellular chx10 turning neurons) [Huerta-Ocampo et al 2021].
If wall-following uses the midbrain orientation circuitry, then a combination of OT and Ppt is plausible following the model on the right. Note, though, that the sustain may not be only Ppt, but could also include other anterior hindbrain systems like R1.a, V.rn (serotonin Raphé nuclei), and possibly R.ip (interpeduncular nucleus), because all of these are associated with brainstem sustained attention network [Alves et al 2022].
S.nr tri-value logic
S.nr (substantia nigra pars reticulata) is a key player in commitment. S.nr provides tonic suppression over essentially every voluntary action. For this essay, consider S.nr as a tri-value logic. The tonic, middle level of S.nr allows ongoing actions to continue, but inhibits starting a new action. A low S.nr value, the classic disinhibition model, allows new actions to start. A high S.nr value stops ongoing actions. The tonic level itself might be adjustable. For decision-making, this tri-value S.nr can support the commitment requirements of sustain, timeout (Stop), and lockout (passive inhibition with overriding Go).
In the diagram above, the tonic S.nr inhibits new actions, but ongoing actions would continue, or a sufficiently strong sense input could start an action. An explicit Go signal would allow a weak sensory input to start an action. An explicit Stop signal would stop all action, regardless of the sense strength. The adjustable tonic level can be influenced by sleep and wake pressure, since S.nr.m is associated with sleep [Liu D et al 2020]. As the animal grows tired, a higher tonic S.nr would discourage new actions and encourage stopping of sustained actions, but the sleep pressure would not outright prevent action.
In the context of decision commitment, disabling S.nr eliminates orientation selectivity: mice are unable to resist orienting to any object in the whisker field [Redgrave et al 1999]. In PD (Parkinson’s disease) an overly-active S.nr produces bradykinesia (slow movements) and akinesia (lack of voluntary movement), and inhibiting S.nr can reduce akinesia and bradykinesia [Hu Y et al 2023], [Lin C et al 2024]. However, an underactive S.nr can produce dyskinesia (twisted postures) where opposing actions activate at the same time, and stimulating S.nr can reduce dyskinesia [Hu Y et al 2023].
Action sustain and timeout
An immediate consequence of commitment is timeout. Without a timeout, an unending commitment can lock an animal into a decision forever. A previous essay already covered a possible timeout circuit using adenosine as a timing neurotransmitter. The S.d2 (striatum with D2 dopamine-receptor) projection neuron have A2a.s (adenosine G-s coupled stimulatory) receptors, and some studies use A2a.s receptors to identify the S.d2 neurons as S.a2a as opposed to the S.d2 convention in the essays. Adenosine builds up around active neurons, partially produced by astrocytes that monitor glutamate activity [Ma et al 2022]. This adenosine progressively activates S.d2 neurons, which stops action using the indirect path.
The above diagram shows a possible timeout circuit for thigmotaxis following a left wall. During the left wall-following, an efference copy via T.pf (parafascicular thalamus) drives glutamate to S.d2 in the striatum. Sustained glutamate in S.d2 produces adenosine, which progressively activates S.d2, which then inhibits the current action using the indirect path of P.ge (external globus pallidus) to H.stn (subthalamic nucleus) to S.nr to stop the left action.
Note that the P.ge / H.stn circuit is complex and oscillatory. This path isn’t necessarily a straight chain as the above diagram would suggest. For example, S.d2 to H.stn / P.ge can switch the mode from irregular, unsynchronized firing to a regular oscillation [Terman et al 2002], or switch from a gamma (~80Hz) to a beta (~20Hz) frequency [Wang Y et al 2024].
Interrupting sustained action
Sometimes sustained actions need to be interrupted, either for dramatic reasons like a predator attack or more mundane situations like stubbing a toe. These interrupts need to be fully general, halting the current action, no matter which action path happens to be active. Note the similarity to sleep, where sleep needs to halt any action.
Two ways of halting action are either a direct halt signal or a deadman’s switch. In a deadman’s switch, a tonic signal maintains normal behavior, and the absence of the signal stops the action. The brain uses this pattern with several instances of high-affinity Gi (G-protein coupled inhibitory) receptors that saturate in normal, tonic activity, but disengage when the neurotransmitter drops. In particular the D2.i (dopamine G-i inhibitory) receptor is high affinity, quickly saturating, that is fully active at normal tonic levels of dopamine and only shuts off when dopamine levels drop.
The diagram above adds a dopamine deadman’s switch to the timeout circuit. Tonic dopamine from V.da (midbrain dopamine) normally inhibits S.d2, allowing for a normal timeout. Because D2.i is a high affinity receptor, a low tonic level of dopamine activates it and quickly saturates the receptor. When dopamine drops below a threshold, the D2.i receptor will deactivate and disinhibit the S.d2 neuron, which rapidly fires the timeout using the indirect path, stopping the action. Dopamine will drop if V.rmtg (rostromedial tegmental) activates. V.rmtg is activated by pain or itch sensations and by many more general failure or disappointment systems.
Lockout and the Sprague effect
The commitment phase needs to lockout alternative distractors. I haven’t found any research on this specific scenario. Decision research generally studies artificial forced-choice scenarios, where each choice is separated by several seconds from another choice, and is forced by a single decision point, like a T-maze or Y-maze, or turning left or right from a central cue port. The design of the typical experiment removes the scenario of sequential, continuous choices. Because of the lack of direct studies, the following discussion is more speculative. Attention is a related, but distinct research area to decision-making. Sustained attention is similar to this commitment issue.
The Sprague effect is related to OT attention. In mammals OT receives excitatory input from C.vis (visual cortex). If the left C.vis is lesioned, the animal will ignore items in contralateral, right visual field. Paradoxically, a following lesion to contralateral, right OT will restore attention to the left visual field [Gambrill et al 2018], [Gebhardt et al 2019], [Jiang et al 2003], [Krauzlis et al 2013]. Further studies have shown this effect with the second lesion to the tectal commissure [Gambrill et al 2018] or to the specific area of contralateral S.nr [Krauzlis et al 2013] or to the entire contralateral Ppt [Valero-Cabré et al 2020].
In frogs there is a direct OT to contralateral OT connection. Unilateral OT legion impairs bilateral visual behavior regardless of looming direction [Gambrill et al 2018]. In contrast unilateral OT lesion deficit in behavior only in lesioned hemifield [Gambrill et al 2018].
This above diagram shows a potential circuit for the Sprague effect. (For simplicity, straightening out crossed output to the motor.) Once a decision to follow a right wall has been made, the ongoing motor action sends an efferent copy to Ppt [Caggiano et al 2018], which projects to S.nr [Durmer and Rosenquist 2001], which inhibits the contralateral OT.d [Durmer and Rosenquist 2001]. Similarly, a motor efferent copy to Ppt also projects to the ipsilateral OT.d [Valero-Cabré et al 2020], which enhances attention to continue following the right wall.
This Ppt sustained attention circuit is similar to the R.is (nucleus isthmus / parabigeminal) circuit in fish [Henriques et al 2019] and birds [Marín et al 2007], covered in essay 19. Like Ppt, R.is has both ACh and GABA components, although in Ppt the components are mixed salt-and-pepper, while R.is has distinct nuclei. Ppt and R.is are sibling areas, both generated from the same progenitors in R1 (hindbrain rhombomere 1), but one is generated before the other [Morello et al 2020]. R.is is better understood because it has simpler connectivity than Ppt. When zebrafish hunt paramecia, R.is sustains attention to a target prey and inhibits attention to other visual areas [Henriques et al 2019]. R.is provides a similar effect in birds [Knudsen 2011], [Marín et al 2007], [Mysore and Knudsen 2011], [Reynaert et al 2023].
Although Ppt has much more complicated connectivity and function, like R.is, it has reciprocal connectivity with OT. Ppt is also active during actions, and it highly heterogenous, and connected with much of the hindbrain motor, both R.pn (pons, anterior hindbrain) and R.my (medulla, central and posterior hindbrain). Like R.is, Ppt proves ACh attention to OT [Isa et al 2021], [Mena-Segovia et al 2008], [Mena-Segovia et al 2017], [Krauzlis et al 2013], [Wolf et al 2015] and as the Sprague studies show, it inhibits the contralateral OT via S.nr.
At the time of choice, many Ppt reflect previous action and outcome. Ppt lesions reduce influence of recent experience on action selection. The Ppt ACh input to OT is possible as a Bayesian prior [Thompson et al 2016].
Passive lockout
An alternative to an active of alternative actions is a passive lockout, which inhibits actions without needing input from a sustain system. Once an action commits, the passive lockout prevents new action. A possible passive lockout involves the H.stn / P.ge pair, which is hyperactive in PD. Akinesia like PD is exactly what’s needed for passive lockout.
In the above circuit, H.stn and P.ge form a spontaneously oscillating circuit at beta frequencies. In PD, this circuit is hyperactive, oscillating at beta frequencies, providing broad movement inhibition [Fischer et al 2017]. This circuit drives S.nr, which inhibits the action.
This description vastly oversimplifies the P.ge / H.stn circuit. The P.ge / H.stn circuit can operate in at least two modes: inhibitory at beta frequencies (H.stn exciting S.nr), and excitatory at gamma (P.ge inhibiting S.nr) [Fisher et al 2017], [Terman et al 2002]. H.stn also has distinct subregions, with H.stn.vm (venture-medial) as almost an extension of H.l [Haynes and Haber 2013], while H.stn.l as distinct functionality [Baunez and Lardeux 2011], [Pasquereau and Turner 2017].
Studies seem to divide on whether H.stn is suitable for a commitment function. H.stn activity terminates at onset of movement [Espinosa-Parrilla et al 2013], which would argue against passive lockout. H.stn gamma increases during movement for humans [Fischer et al 2017], but others point out that H.stn beta are brief bursts, not sustained [Feingold et al 2015], and H.stn beta in humans is active for acute stopping [Wessel et al 2016].
A second passive lockout is in the striatum itself, discouraging new actions by default. S.pn (striatum projection neurons), both S.d1 (D1 receptor S.pn) and S.d2, are hyper polarized, making them harder to drive than most neurons. Secondarily, new actions are inhibited by the feedforward, fast-spiking S.pv (parvalbumin) neurons, which inhibit S.d1 and S.d2 before they can be activated. S.pv activates before S.d1 and S.d2 [Gage et al 2010], [Lee C et al 2019], [O’Hare et al 2017], [Yim et al 2011].
This S.pv inhibition is suppressed by sustained action using a retrograde eCB (endocannabinoid) system that disinhibits both S.d1 and S.d2 by inhibiting GABA release from S.pv [Narushima et al 2006], [Adermark et al 2009], [Mathur and Lovinger 2012].
Active initialization
A passive lockout system needs to be paired with an active initialization. If new actions are passively inhibited by default, a new action needs extra effort to cross the barrier. Possible active initialization nodes include OT, Ppt, as well as the S.d1 direct path.
The following diagram shows a possible active initialization. The passive lockout subcircuit is the same as before. The active initialization would logically use the S.d1 path. Phasic dopamine activates the Go path, both by enabling the S.d1 input and their output, because D1.s receptors are on inputs to S.d1 and on the axons in S.nr, which enables the direct path to disinhibit the S.nr.
This Go circuit is for wall-following, which uses the lateral-line as a wall distance sensor. The lateral line sense is input to the OT orientation circuit, which excites both the S.d1 path via T.pf and the V.da path, which will add a phasic DA burst to enhance the S.d1 circuit, giving it an extra boost to overcome the barriers.
Note that OT / Ppt also inhibits contralateral OT as described in the Sprague effect system, and the OT to V.da excitation is ipsilateral, but OT also inhibits the contralateral V.da via V.rmtg [Pradel et al 2021]. So this circuit is also part of the active lockout system.
Consider the striatum passive lockout circuit again, which discouraged new actions but enabled sustained actions. That passive lockout implies the necessity of an extra push for new actions. Phasic dopamine bursts could provide that extra push. A burst of dopamine activates the low-affinity D1.s receptors in S.d1, allowing S.d1 to override its intrinsic hyperpolarization and the feedforward S.pv inhibition and initiate a new action.
Pretectum suppression of OT
This essay uses the aquatic-only lateral-line for thigmotaxis. Thigmotaxis is an interesting system because the animal must be weakly attracted to the wall but simultaneously repelled by the wall to avoid collision. M.pt (pretectum) is an obstacle avoidance system in the midbrain and OT is an orienting system. In non-mammalian vertebrates (reptiles, birds, frogs), the striatum projects directly to M.pt, which inhibits OT [Krauzlis et al 2018]. If the animal gets too close to the wall, M.pt should inhibit the OT orientation and avoid the wall, but if the animal is far enough from the wall, it should approach the wall with thigmotaxis, suppressing the avoidance circuit.
The above circuit shows this potential thigmotaxis circuit. Normally, M.pt avoids the wall and suppresses OT to keep any OT orientation from running into the wall. But during thigmotaxis, the S circuit will suppress M.pt to allow the animal to get closer to the wall.
Simulation
The simulation divides thigmotaxis into several systems. An obstacle system roughly corresponds to M.pt and keeps the animal from running into a wall. An orientation system provides an attractive drive toward the wall. These two systems are now designed as independent and general, where sensory input is external. For example, the lateral line drives both the obstacle and orient systems, but the code for obstacle and orient systems are ignorant of the lateral line itself.
The decision commitment uses a loop with a sustain module and a striatum module. The sustain roughly corresponds to Ppt with possible associated areas like V.dr and R1.a, because the simulation is more abstract than directly implementing each neural ganglia. The striatum module provides with timeout function with an adenosine-lie timeout. The sustain also provides an active inhibitory lockout function, following the Sprague effect studies.
References
Adermark L, Talani G, Lovinger DM. Endocannabinoid-dependent plasticity at GABAergic and glutamatergic synapses in the striatum is regulated by synaptic activity. Eur J Neurosci. 2009 Jan;29(1):32-41.
Alves PN, Forkel SJ, Corbetta M, Thiebaut de Schotten M. The subcortical and neurochemical organization of the ventral and dorsal attention networks. Commun Biol. 2022 Dec 7;5(1):1343.
Baunez C, Lardeux S. Frontal cortex-like functions of the subthalamic nucleus. Front Syst Neurosci. 2011 Oct 11;5:83.
Caggiano V, Leiras R, Goñi-Erro H, Masini D, Bellardita C, Bouvier J, Caldeira V, Fisone G, Kiehn O. Midbrain circuits that set locomotor speed and gait selection. Nature. 2018 Jan 25;553(7689):455-460.
Durmer JS, Rosenquist AC. Ibotenic acid lesions in the pedunculopontine region result in recovery of visual orienting in the hemianopic cat. Neuroscience. 2001;106(4):765-81.
Espinosa-Parrilla JF, Baunez C, Apicella P. Linking reward processing to behavioral output: motor and motivational integration in the primate subthalamic nucleus. Front Comput Neurosci. 2013 Dec 17;7:175.
Feingold J, Gibson DJ, DePasquale B, Graybiel AM. Bursts of beta oscillation differentiate postperformance activity in the striatum and motor cortex of monkeys performing movement tasks. Proc Natl Acad Sci U S A. 2015 Nov 3;112(44):13687-92.
Fischer P, Pogosyan A, Herz DM, Cheeran B, Green AL, Fitzgerald J, Aziz TZ, Hyam J, Little S, Foltynie T, Limousin P, Zrinzo L, Brown P, Tan H. Subthalamic nucleus gamma activity increases not only during movement but also during movement inhibition. Elife. 2017 Jul 25;6:e23947.
Gage GJ, Stoetzner CR, Wiltschko AB, Berke JD. Selective activation of striatal fast-spiking interneurons during choice execution. Neuron. 2010 Aug 12;67(3):466-79.
Gambrill AC, Faulkner RL, Cline HT. Direct intertectal inputs are an integral component of the bilateral sensorimotor circuit for behavior in Xenopus tadpoles. J Neurophysiol. 2018 May 1;119(5):1947-1961.
Gebhardt C, Auer TO, Henriques PM, Rajan G, Duroure K, Bianco IH, Del Bene F. An interhemispheric neural circuit allowing binocular integration in the optic tectum. Nat Commun. 2019 Nov 29;10(1):5471.
Ghitani N, Bayguinov PO, Basso MA, Jackson MB. A sodium afterdepolarization in rat superior colliculus neurons and its contribution to population activity. J Neurophysiol. 2016 Jul 1;116(1):191-200.
Haynes WI, Haber SN. The organization of prefrontal-subthalamic inputs in primates provides an anatomical substrate for both functional specificity and integration: implications for Basal Ganglia models and deep brain stimulation. J Neurosci. 2013 Mar 13;33(11):4804-14.
Henriques, P.M., Rahman, N., Jackson, S.E. and Bianco, I.H., 2019. Nucleus isthmi is required to sustain target pursuit during visually guided prey-catching. Current Biology, 29(11), pp.1771-1786.
Hu, Y., Ma, T.C., Alberico, S.L., Ding, Y., Jin, L. and Kang, U.J., 2023. Substantia Nigra pars reticulata projections to the pedunculopontine nucleus modulate dyskinesia. Movement Disorders, 38(10), pp.1850-1860.
Huerta-Ocampo I, Dautan D, Gut NK, Khan B, Mena-Segovia J. Whole-brain mapping of monosynaptic inputs to midbrain cholinergic neurons. Sci Rep. 2021 Apr 27;11(1):9055.
Isa T, Marquez-Legorreta E, Grillner S, Scott EK. The tectum/superior colliculus as the vertebrate solution for spatial sensory integration and action. Curr Biol. 2021 Jun 7;31(11):R741-R762.
Jiang H, Stein BE, McHaffie JG. Opposing basal ganglia processes shape midbrain visuomotor activity bilaterally. Nature. 2003 Jun 26;423(6943):982-6.
Knudsen EI. Evolution of neural processing for visual perception in vertebrates. J Comp Neurol. 2020 Dec 1;528(17):2888-2901.
Krauzlis RJ, Lovejoy LP, Zénon A. Superior colliculus and visual spatial attention. Annu Rev Neurosci. 2013 Jul 8;36:165-82.
Krauzlis RJ, Bogadhi AR, Herman JP, Bollimunta A. Selective attention without a neocortex. Cortex. 2018 May;102:161-175.
Lee, C. R., Yonk, A. J., Wiskerke, J., Paradiso, K. G., Tepper, J. M., & Margolis, D. J. (2019). Opposing influence of sensory and motor cortical input on striatal circuitry and choice behavior. Current Biology, 29(8), 1313-1323.
Lin, C., Ridder, M., Zhong, J., Albornoz, E.A., Sedlak, P., Xu, L., Woodruff, T.M., Chen, F. and Sah, P., 2024. Modulation of pedunculopontine input to the basal ganglia relieves motor symptoms in Parkinsonian mice. bioRxiv, pp.2024-03.
Lintz MJ, Essig J, Zylberberg J, Felsen G. Spatial representations in the superior colliculus are modulated by competition among targets. Neuroscience. 2019 Jun 1;408:191-203.
Liu, D., Li, W., Ma, C., Zheng, W., Yao, Y., Tso, C. F., … & Dan, Y. (2020). A common hub for sleep and motor control in the substantia nigra. Science, 367(6476), 440-445.
Ma L, Day-Cooney J, Benavides OJ, Muniak MA, Qin M, Ding JB, Mao T, Zhong H. Locomotion activates PKA through dopamine and adenosine in striatal neurons. Nature. 2022 Nov;611(7937):762-768.
Marín, G., Salas, C., Sentis, E., Rojas, X., Letelier, J.C. and Mpodozis, J., 2007. A cholinergic gating mechanism controlled by competitive interactions in the optic tectum of the pigeon. Journal of Neuroscience, 27(30), pp.8112-8121.
Mathur BN, Lovinger DM. Endocannabinoid-dopamine interactions in striatal synaptic plasticity. Front Pharmacol. 2012 Apr 19;3:66.
Mena-Segovia J, Sims HM, Magill PJ, Bolam JP. Cholinergic brainstem neurons modulate cortical gamma activity during slow oscillations. J Physiol. 2008 Jun 15;586(12):2947-60.
Mena-Segovia, Juan, and J. Paul Bolam. Rethinking the pedunculopontine nucleus: from cellular organization to function. Neuron 94.1 (2017): 7-18.
Morello F, Borshagovski D, Survila M, Tikker L, Sadik-Ogli S, Kirjavainen A, Estartús N, Knaapi L, Lahti L, Törönen P, Mazutis L, Delogu A, Salminen M, Achim K, Partanen J. Molecular Fingerprint and Developmental Regulation of the Tegmental GABAergic and Glutamatergic Neurons Derived from the Anterior Hindbrain. Cell Rep. 2020 Oct 13;33(2):108268.
Munoz DP, Wurtz RH. Saccade-related activity in monkey superior colliculus. I. Characteristics of burst and buildup cells. J Neurophysiol. 1995 Jun;73(6):2313-33.
Mysore SP, Knudsen EI. The role of a midbrain network in competitive stimulus selection. Curr Opin Neurobiol. 2011 Aug;21(4):653-60.
Narushima M, Uchigashima M, Hashimoto K, Watanabe M, Kano M. Depolarization-induced suppression of inhibition mediated by endocannabinoids at synapses from fast-spiking interneurons to medium spiny neurons in the striatum. Eur J Neurosci. 2006 Oct;24(8):2246-52.
O’Hare JK, Li H, Kim N, Gaidis E, Ade K, Beck J, Yin H, Calakos N. Striatal fast-spiking interneurons selectively modulate circuit output and are required for habitual behavior. Elife. 2017 Sep 5;6:e26231.
Pasquereau B, Turner RS. A selective role for ventromedial subthalamic nucleus in inhibitory control. Elife. 2017 Dec 4;6:e31627.
Pradel K, Drwiȩga G, Błasiak T. Superior Colliculus Controls the Activity of the Rostromedial Tegmental Nuclei in an Asymmetrical Manner. J Neurosci. 2021 May 5;41(18):4006-4022.
Redgrave, Peter, Tony J. Prescott, and Kevin Gurney. The basal ganglia: a vertebrate solution to the selection problem?. Neuroscience 89.4 (1999): 1009-1023.
Reynaert B, Morales C, Mpodozis J, Letelier JC, Marín GJ. A blinking focal pattern of re-entrant activity in the avian tectum. Curr Biol. 2023 Jan 9;33(1):1-14.e4.
Schall JD. Accumulators, Neurons, and Response Time. Trends Neurosci. 2019 Dec;42(12):848-860. doi: 10.1016/j.tins.2019.10.001.
Stine GM, Trautmann EM, Jeurissen D, Shadlen MN. A neural mechanism for terminating decisions. Neuron. 2023 Aug 16;111(16):2601-2613.e5.
Terman D, Rubin JE, Yew AC, Wilson CJ. Activity patterns in a model for the subthalamopallidal network of the basal ganglia. J Neurosci. 2002 Apr 1;22(7):2963-76.
Thompson JA, Costabile JD, Felsen G. Mesencephalic representations of recent experience influence decision making. Elife. 2016 Jul 25;5:e16572.
Thompson AC, Aizenman CD. Characterization of Na+ currents regulating intrinsic excitability of optic tectal neurons. Life Sci Alliance. 2023 Nov 2;7(1):e202302232.
Valero-Cabré A, Toba MN, Hilgetag CC, Rushmore RJ. Perturbation-driven paradoxical facilitation of visuo-spatial function: Revisiting the ‘Sprague effect’. Cortex. 2020 Jan;122:10-39.
Wang Y, Wang L, Manssuer L, Zhao YJ, Ding Q, Pan Y, Huang P, Li D, Voon V. Subthalamic stimulation causally modulates human voluntary decision-making to stay or go. NPJ Parkinsons Dis. 2024 Nov 2;10(1):210.
Wessel JR, Jenkinson N, Brittain JS, Voets SH, Aziz TZ, Aron AR. Surprise disrupts cognition via a fronto-basal ganglia suppressive mechanism. Nat Commun. 2016 Apr 18;7:11195.
Wolf AB, Lintz MJ, Costabile JD, Thompson JA, Stubblefield EA, Felsen G. An integrative role for the superior colliculus in selecting targets for movements. J Neurophysiol. 2015 Oct;114(4):2118-31.
Yim, M. Y., Aertsen, A., & Kumar, A. (2011). Significance of input correlations in striatal function. PLoS Computational Biology, 7(11), e1002254.
Zeymer M, von der Emde G, Wullimann MF. The Mormyrid Optic Tectum Is a Topographic Interface for Active Electrolocation and Visual Sensing. Front Neuroanat. 2018 Oct 1;12:79. doi: 10.3389/fnana.2018.00079.
