Neuron

Previews of manipulating the circuit at the moment in time when phasic DA is elevated. Finally, to support their contention that the behavioral effects were the result of falsely signaling the presence or absence of the reward, the authors developed tasks in which they physically increased or decreased the presence or absence of reward without stimulating any brain region. They found that omitting all risky lever reward led to a decrease in choice of that lever, similar to LHb stimulation during delivery of the risky reward. Similarly, omitting all safe lever reward led to a switch in preference for the risky lever, similar to LHb stimulation during delivery of the safe reward. To mimic VTA stimulation, the authors increased the odds of receiving reward after a risky press to 100%, which likewise increased choice of the risky lever. Thus, the act of physically omitting or presenting reward had the same impact on behavior as stimulation of the brain regions thought to be responsible for tracking reward omissions and presentations. In total, Stopper et al. (2014) have put together a convincing set of experiments that highlight the critically important role phasic DA plays in modulating risky

choice. As with any experiment, the results leave one with many questions. Where are the downstream targets of this information? Could these findings be replicated if one optogenetically modulated DAergic terminals in separate regions known to impact risky decisionmaking, such as the nucleus accumbens, amygdala, or prefrontal cortex (Cardinal and Howes, 2005; Ghods-Sharifi et al., 2009; Mobini et al., 2002)? In addition to risk, what is the role of the LHb-RMTgVTA circuit in decision-making at large? The same group of authors recently found that pharmacological manipulation of this circuit could alter both effort and delay decision-making (Stopper and Floresco, 2014). It certainly seems possible that phasic DA could impact these behaviors as well. Investigation of this and other circuitry could help elucidate which processes are important for broad cost/ benefit analysis, and which are specific to particular modalities such as risk, delay, or effort.

Fiorillo, C.D., Tobler, P.N., and Schultz, W. (2003). Science 299, 1898–1902. Ghods-Sharifi, S., St Onge, J.R., and Floresco, S.B. (2009). J. Neurosci. 29, 5251–5259. Hikosaka, O., Sesack, S.R., Lecourtier, L., and Shepard, P.D. (2008). J. Neurosci. 28, 11825– 11829. Jhou, T.C., Geisler, S., Marinelli, M., Degarmo, B.A., and Zahm, D.S. (2009). J. Comp. Neurol. 513, 566–596. Ji, H., and Shepard, P.D. (2007). J. Neurosci. 27, 6923–6930. Mobini, S., Body, S., Ho, M.Y., Bradshaw, C.M., Szabadi, E., Deakin, J.F., and Anderson, I.M. (2002). Psychopharmacology (Berl.) 160, 290–298. Schultz, W. (1998). J. Neurophysiol. 80, 1–27. St Onge, J.R., Chiu, Y.C., and Floresco, S.B. (2010). Psychopharmacology (Berl.) 211, 209–221. Stamatakis, A.M., and Stuber, G.D. (2012). Nat. Neurosci. 15, 1105–1107. Stopper, C.M., and Floresco, S.B. (2014). Nat. Neurosci. 17, 33–35. Stopper, C.M., Khayambashi, S., and Floresco, S.B. (2012). Neuropsychopharmacology 38, 715–728.

REFERENCES

Stopper, C.M., Tse, M.T.L., Montes, D.R., Wiedman, C.R., and Floresco, S.B. (2014). Neuron 84, this issue, 177–189.

Cardinal, R.N., and Howes, N.J. (2005). BMC Neurosci. 6, 37.

Sugam, J.A., Day, J.J., Wightman, R.M., and Carelli, R.M. (2012). Biol. Psychiatry 71, 199–205.

A Cortical Rein on the Tectum’s Gain Sylvia Schro¨der1,* and Matteo Carandini1 1University College London, 11-43 Bath Street, London, EC1V 9EL, UK *Correspondence: [email protected] http://dx.doi.org/10.1016/j.neuron.2014.09.021

The superior colliculus, or tectum, is a key sensorimotor structure that long predates the cortex. In this issue of Neuron, Zhao et al. (2014) show that the visual cortex controls the tectum’s gain precisely and retinotopically, without otherwise altering its operations. You are intent on typing a document, and an alert pops up in a corner of your screen. Before you realize it, your eyes are already on it, scanning its text. Most likely, this fast and automatic reaction was mediated by your superior colliculus, or tectum. The superior colliculus (SC) is a brain structure of remarkable organization and

effectiveness, aimed at integrating sensory inputs to produce motor outputs. Stacked one above the other, its layers contain maps that go from sensory to motor, aligned to each other to highlight locations of interest and move eyes, head, and body toward them. Its origins long precede the cerebral cortex—in nonmammals it is called the optic tectum—

6 Neuron 84, October 1, 2014 ª2014 Elsevier Inc.

and it has maintained a strategic position even as the cerebral cortex has grown to cover it (Schiller, 2011). Indeed, the cortex seems to take great care to influence the SC, sending it axons from a wide array of cortical areas (Figure 1A). Axons from visual cortex tend to target the more superficial layers, which are visual, while those from

Neuron

Previews auditory, somatosensory, and course of responses. Impormotor cortices reach deeper tantly, these results were oblayers, which are auditory, tained in awake mice. When tactile, and motor, and often the experiments were repeated multisensory. The orderly and in anesthetized mice, SC repervasive pattern of these sponses were unaffected by cortical inputs was observed in V1 inactivation. What could be the cat (Harting et al., 1992), mediating these effects? and later in the mouse, thanks A simple possibility, which is to the Allen Mouse Connectivity unlikely, is that V1 inactivation Atlas (Q. Wang et al., 2013, Soc. caused a global change in brain Neurosci., abstract 488.04). state. The authors could rule out But what does the cortex tell this possibility thanks to the the SC? This question has precision of optogenetics, been tackled extensively since whereby inactivation can be the 1960s, with contradictory spatially restricted. Zhao et al. conclusions. A key body of re(2014) achieved this by exsults concerns the superficial pressing Channelrhodopsin-2 layers, which are clearly visual, in a small region of V1 and found and receive direct visual signals that the effect of inactivation from the retina, as well as prowas strictly retinotopic: inactijections from primary visual vation of the V1 region only Figure 1. Probing Visual Inputs to the Superior Colliculus cortex (V1). How do their reaffected tectal neurons that (A) The superior colliculus (SC) receives retinotopic visual inputs from sponses change when their V1 represent the same portion of retina, primary visual cortex (V1), higher areas of the cortex, and basal inputs are removed? the visual field. The effect of V1 ganglia (open arrow denotes inhibition, closed arrows denote excitation). Until the end of the last cenon SC responsiveness, there(B) Time course of four visual stimuli: spots growing in size at various speeds (from black to red, fastest to slowest). The vertical axis represents tury, the most refined method fore, is local and orderly mapspatial extent, from 20 to 20 of visual angle, relative to the receptive to inactivate a region of cortex ped, not global. field center. was to cool it. Multiple studies Another simple possibility, (C) Responses evoked by these stimuli in superficial layers of the SC. (D) Those same responses, when V1 is optogenetically inactivated. also unlikely, is that these efemployed this technique and (E) Effect of V1 inactivation on the peak SC responses to those four stimuli. fects reveal a direct contribumeasured its effects on visual (F) Responses of layer 5 neurons in V1 to the same four stimuli. Responses tion of V1 visual responses to responses in the superficial in (B)–(F) are highly simplified depictions of the actual data obtained by SC activity. This is unlikely, SC layers. Some studies perZhao et al. (2014). because V1 responses to loomformed in cats found changes ing stimuli were remarkably in visual preferences accompanied, in general but not exclusively, by re- sin-2 in a majority of inhibitory neurons different from those measured in SC ductions in responsiveness (Ogasawara (those that are Parvalbumin positive) in a (Figure 1F). Whereas the peak latency of et al., 1984; Wickelgren and Sterling, local region of area V1 and activated SC neurons showed little variation with 1969). Measurements in monkeys, these interneurons by shining a blue light speed, neurons in V1 reached peak instead, found no effect on visual re- on them. At the same time, they recorded firing whenever the stimulus reached the sponses in the superficial layers; these in the upper layers of superior colliculus, preferred size. These differences in time responses were reduced only in deeper in response to simple visual stimuli: dots course between V1 and SC suggest that that became progressively larger over V1 activity is not simply added to or multilayers (Schiller et al., 1974). One possibility for these conflicting re- time, looming toward the animal plied with the SC responses. Perhaps these results could be reconsults is that cortical control of SC differs (Figure 1B). SC responded strongly to across species. Another possibility, how- these stimuli, especially when their size ciled if one could record specifically ever, is that cooling is not sufficiently grew fairly quickly (Figure 1C). Inactiva- from the V1 neurons in layer 5 that project reliable to investigate these effects. More- tion of V1 markedly reduced these re- to the SC. These neurons show some difover, all of these studies were performed sponses, but it left their time course and ferences in visual properties relative to the in anesthetized animals, and SC may stimulus dependence remarkably unaf- rest of layer 5 (Palmer and Rosenquist, work very differently under anesthesia fected (Figure 1D). The only effect on the 1974; Wurtz and Albano, 1980). If their responses was to reduce them to approx- response time course was more similar (Schiller, 2011). An elegant study by Zhao et al. (2014) in imately half of their original size to that of SC neurons, a direct interaction this issue of Neuron exploits the powerful (Figure 1E). Thus, the cortex provided would seem more plausible. Future exoptogenetic techniques now available in facilitative input to the neurons in the up- periments may shed light on this matter mice to overcome these limitations. To per layers of SC, and the effect is purely by specifically targeting corticotectal neusilence a targeted cortical region, they one of changing responsiveness, not the rons and comparing their properties to the used mice that express Channelrhodop- visual preferences, and not the time rest of the neural population. Neuron 84, October 1, 2014 ª2014 Elsevier Inc. 7

Neuron

Previews An alternative and perhaps more likely explanation for the results lies in the competitive, gain-control circuits that are intrinsic to SC. There is substantial evidence for competitive lateral circuits in the SC, both in the intermediate layers (Munoz and Istvan, 1998) and in superficial layers (Vokoun et al., 2014). These competitive circuits can also operate across the intertectal commissure, which joins the left and right SC. They control response gain in a precise, multiplicative manner (Vokoun et al., 2014). Because of these competitive circuits, even a small change in the local input received by a region of SC may be amplified into a major advantage or disadvantage in that region’s competition with the rest of the SC. This amplification may lead to a threshold effect: cortical activity might enhance SC responses only if it crosses a given threshold. This threshold might be so low that even spontaneous activity crosses it, unless it is reduced by optogenetic inactivation. Or visual responses might be required, but their precise time course might be unimportant, as long as V1 activity exceeds the threshold. This hypothesis would be consistent with some of the classic studies that performed ablation or cooling of visual cortex. Ablating V1 leads to a reduction in SC response, as removal of cortical input leaves each colliculus with the suppression caused by the contralateral colliculus. Cutting the connection between the colliculi then reinstates their responsiveness (Lomber and Payne, 1996; Sprague, 1966).

In fact, the effects of V1 on SC are not necessarily direct. They may be mediated by any retinotopically organized brain region that is under the influence of V1 input and provides output to the SC (Figure 1A). Obvious candidates are higher visual cortical areas. They receive excitatory input from V1 and have excitatory projections to SC. In addition, SC may receive retinotopic, tonic inhibition from the basal ganglia (as has been observed in monkeys [Hikosaka and Wurtz, 1983]). V1 may in turn reduce this inhibition, thus disinhibiting SC (Figure 1A). These questions of mechanism are accompanied by questions of function: does V1 control SC during natural vision and behavior? It is likely that the large effects on SC revealed by complete inactivation of a local region of V1 are extreme cases of modulations that happen throughout normal vision and behavior. Characterizing the nature and magnitude of these changes and their functional significance would be a very interesting next step. Meanwhile, the results revealed by Zhao and colleagues (2014) stand out for their elegance and usefulness. The method of optogenetic inactivation is much more precise and quickly reversible than cooling, allowing one to alternate rapidly trials with inactivation and control trials. The results obtained with this technique also stand out for their simplicity. They reveal that the cortex can essentially double the responsiveness of SC. This is a powerful influence, but one that is exerted delicately, without interfering with either the time course of responses or the visual preferences seen in the tectum.

8 Neuron 84, October 1, 2014 ª2014 Elsevier Inc.

ACKNOWLEDGMENTS We thank Michele Basso, Robert McPeek, Massimo Scanziani, and Nicholas Steinmetz for helpful discussions. S.S. is supported by the People Programme (Marie Curie Actions) of the European Union. M.C. holds the GlaxoSmithKline / Fight for Sight Chair in Visual Neuroscience. REFERENCES Harting, J.K., Updyke, B.V., and Van Lieshout, D.P. (1992). J. Comp. Neurol. 324, 379–414. Hikosaka, O., and Wurtz, J. Neurophysiol. 49, 1285–1301.

R.H.

(1983).

Lomber, S.G., and Payne, B.R. (1996). Vis. Neurosci. 13, 1143–1156. Munoz, D.P., and Istvan, J. Neurophysiol. 79, 1193–1209.

P.J.

(1998).

Ogasawara, K., McHaffie, J.G., and Stein, B.E. (1984). J. Neurophysiol. 52, 1226–1245. Palmer, L.A., and Rosenquist, A.C. (1974). Brain Res. 67, 27–42. Schiller, P.H. (2011). The superior colliculus and visual function. In Compr. Physiol. 2011, Supplement 3: Handbook of Physiology, The Nervous System, Sensory Processes. 457–505. First published in print 1984. http://dx.doi.org/10.1002/ cphy.cp010311. Schiller, P.H., Stryker, M., Cynader, M., and Berman, N. (1974). J. Neurophysiol. 37, 181–194. Sprague, J.M. (1966). Science 153, 1544–1547. Vokoun, C.R., Huang, X., Jackson, M.B., and Basso, M.A. (2014). J. Neurosci. 34, 7976– 7987. Wickelgren, B.G., and Sterling, J. Neurophysiol. 32, 16–23.

P.

(1969).

Wurtz, R.H., and Albano, J.E. (1980). Annu. Rev. Neurosci. 3, 189–226. Zhao, X., Liu, M., and Cang, J. (2014). Neuron 84, this issue, 202–213.