DispatchR223

was sufficient to dominantly suppressWldS-mediated axonal protection [4].Both studies therefore showed thatexpression of Nmnat or WldS proteinsfailed to protect axons after injury whenMiro or Milton function wascompromised, indicating that thepresence of mitochondria in the axonis required for Nmnat/Wlds-mediatedaxonal protection (Figure 1). However,as it is unclear whether the transport ofother organelles and proteins is alsodisrupted with decreased Miro/Miltonfunction, it will be critical todifferentiate whether the loss of WldS

protection is due to decreasedmitochondrial numbers and functionin the axon, or simply to decreasedtransport or expression of the WldS

protein in the axon.The two reports together

demonstrate that WldS/Nmnat activityenhances mitochondrial motility andCa2+ buffering and that themitochondrion is an organellenecessary for WldS/Nmnat-mediatedaxonal protection. The processesregulating axon degeneration and theWldS/Nmnat enzymatic activitiesthat are critical for axonal protectionthus converge at axonal mitochondria.A clear future direction is to addresswhether directly enhancing thesemitochondrial functions is sufficientto exert axonal protection. Moreover,identifying whether known enzymaticmetabolites of the WldS/Nmnatproteins, such as NAD+, interact withmolecules in the mitochondria will beinstrumental in understanding thefull downstream mechanisms ofWldS/Nmnat-mediated axon

protection. Although there is still muchto learn about the molecular processesregulating axonal degenerationand survival, these two reports havegiven us a boost by placing the focussquarely on the axonal mitochondria.

References1. Deckwerth, T.L., and Johnson, E.M., Jr. (1994).

Neurites can remain viable after destruction ofthe neuronal soma by programmed cell death(apoptosis). Dev. Biol. 165, 63–72.

2. Wang, J.T., Medress, Z.A., and Barres, B.A.(2012). Axon degeneration: molecularmechanisms of a self-destruction pathway.J. Cell Biol. 196, 7–18.

3. Coleman, M.P., and Freeman, M.R. (2010).Wallerian degeneration, wld(s), and nmnat.Annu. Rev. Neurosci. 33, 245–267.

4. Avery, M.A., Rooney, T., Wishart, T.M.,Pandya, J.D., Gillingwater, T.H., Geddes, J.W.,Sullivan, P., and Freeman, M.R. (2012). WldS

prevents axon degeneration through increasedmitochondrial flux and enhanced mitochondrialCa2+ buffering. Curr. Biol. 22, 596–600.

5. Fang, Y., Soares, L., Teng, X., Geary, M., andBonini, N. (2012). A novel Drosophila model ofnerve injury reveals an essential role of Nmnatin maintaining axonal integrity. Curr. Biol. 22,590–595.

6. Lunn, E.R., Perry, V.H., Brown, M.C., Rosen, H.,and Gordon, S. (1989). Absence of Walleriandegeneration does not hinder regeneration inperipheral nerve. Eur. J. Neurosci. 1, 27–33.

7. Mack, T.G., Reiner, M., Beirowski, B., Mi, W.,Emanuelli, M., Wagner, D., Thomson, D.,Gillingwater, T., Court, F., Conforti, L., et al.(2001). Wallerian degeneration of injured axonsand synapses is delayed by a Ube4b/Nmnatchimeric gene. Nat. Neurosci. 4, 1199–1206.

8. Hoopfer, E.D., McLaughlin, T., Watts, R.J.,Schuldiner, O., O’Leary, D.D., and Luo, L.(2006). Wlds protection distinguishes axondegeneration following injury from naturallyoccurring developmental pruning. Neuron 50,883–895.

9. MacDonald, J.M., Beach, M.G., Porpiglia, E.,Sheehan, A.E., Watts, R.J., and Freeman, M.R.(2006). The Drosophila cell corpse engulfmentreceptor Draper mediates glial clearance ofsevered axons. Neuron 50, 869–881.

10. Beirowski, B., Babetto, E., Gilley, J.,Mazzola, F., Conforti, L., Janeckova, L.,Magni, G., Ribchester, R.R., and Coleman, M.P.(2009). Non-nuclear Wld(S) determines its

neuroprotective efficacy for axons andsynapses in vivo. J. Neurosci. 29, 653–668.

11. Sasaki, Y., Vohra, B.P., Baloh, R.H., andMilbrandt, J. (2009). Transgenic miceexpressing the Nmnat1 protein manifest robustdelay in axonal degeneration in vivo.J. Neurosci. 29, 6526–6534.

12. Avery, M.A., Sheehan, A.E., Kerr, K.S.,Wang, J., and Freeman, M.R. (2009). Wld Srequires Nmnat1 enzymatic activityand N16-VCP interactions to suppressWallerian degeneration. J. Cell Biol. 184,501–513.

13. Conforti, L., Wilbrey, A., Morreale, G.,Janeckova, L., Beirowski, B., Adalbert, R.,Mazzola, F., Di Stefano, M., Hartley, R.,Babetto, E., et al. (2009). Wld S protein requiresNmnat activity and a short N-terminal sequenceto protect axons in mice. J. Cell Biol. 184,491–500.

14. Gilley, J., and Coleman, M.P. (2010).Endogenous Nmnat2 is an essential survivalfactor for maintenance of healthy axons.PLoS Biol. 8, e1000300.

15. Yahata, N., Yuasa, S., and Araki, T. (2009).Nicotinamide mononucleotideadenylyltransferase expression inmitochondrial matrix delays Walleriandegeneration. J. Neurosci. 29, 6276–6284.

16. Wang, X., and Schwarz, T.L. (2009). Themechanism of Ca2+-dependent regulation ofkinesin-mediated mitochondrial motility. Cell136, 163–174.

17. MacAskill, A.F., and Kittler, J.T. (2010). Controlof mitochondrial transport and localization inneurons. Trends Cell Biol. 20, 102–112.

18. Glater, E.E., Megeath, L.J., Stowers, R.S., andSchwarz, T.L. (2006). Axonal transport ofmitochondria requires milton to recruit kinesinheavy chain and is light chain independent.J. Cell Biol. 173, 545–557.

19. Stowers, R.S., Megeath, L.J., Gorska-Andrzejak, J., Meinertzhagen, I.A., andSchwarz, T.L. (2002). Axonal transport ofmitochondria to synapses depends on milton,a novel Drosophila protein. Neuron 36,1063–1077.

Department of Neurobiology, StanfordUniversity School of Medicine, D231 FairchildBuilding, 299 Campus Drive, Stanford,CA 94305, USA.*E-mail: jtw@stanford.edu

DOI: 10.1016/j.cub.2012.02.056

Visual Perception: Knowing What toExpect

If perception is hypothesis, where do the hypotheses come from? A new studysuggests that the human visual system uses the history of past stimulation topredict its current input.

Colin W.G. Clifford

It is often said that we live in a changingworld. As we go through life we adaptto those changes and build upexpectations of what the future willhold. Our sensory systems facea similar challenge in dealing with

different environments. There aremany examples of sensory systemsthat are in some fashion optimised totheir natural environment: consider, forexample, the large eyes of thenocturnal bush baby or the acute senseof smell of the foraging honey bee.Sensory adaptation can be viewed as

a process by which our sensorysystems tend to remain optimized toa changing environment. Under thisview, sensory systems are adaptivesystems perhaps sharing principles ofoperation with systems as diverse asant colonies and economies. In hisclassic book Adaptation in Naturaland Artificial Systems, Holland [1]poses several fundamental questionsfor the study of adaptive systems.‘‘What part of the history of itsinteraction with the environment doesthe organism retain?’’ is of keyimportance as it asks what knowledgedrives the system to adapt. Thisquestion is directly addressed in the

… …… ? Ambiguous test

Pseudo-random sequence of adapting orientations

p( (t0) | (t0-t))

Time

p = 0.5

Current Biology

Figure 1. Dependence of perception on history of past stimulation.

The main finding of Chopin and Mamassian [2] is illustrated schematically. The perceivedorientation of an ambiguous test stimulus depends in a characteristic way on the series oforientations shown in the minutes prior to the test. The test is less likely to be perceived assimilar to orientations seen shortly (up to 3 minutes) before but more likely to be perceivedas similar to orientations seen further into the past (5–10 minutes).

Current Biology Vol 22 No 7R224

context of the human visual system bya study [2] published in this issue ofCurrent Biology.

Chopin and Mamassian [2]conducted two experiments in whichambiguous test stimuli were presentedat regular intervals within a series ofadapting gratings varying randomlybetween two orientations. In the firstexperiment, each test stimulusconsisted of a pair of gratings of thesesame two orientations presented oneto each eye. These two orientationscompete for perceptual dominance,a phenomenon known as binocularrivalry [3]. Subjects reported thedominant percept at each testpresentation. Adaptation typicallygenerates negative aftereffects, suchthat sensitivity to stimuli similar to theadaptor decreases and the perceptionof subsequent test stimuli is repelledaway from the adaptor [4]. Suchnegative effects were observed byChopin and Mamassian [2] atintervals of up to three minutesbetween adaptor and test. At longerintervals, however, the effect wasreversed, such that stimuli presented5–13 minutes previously werepredictive of the perceptualinterpretation of ambiguous visualinformation (Figure 1).

This pattern of dependence ofthe perceived orientation of rivalroustest stimuli on the series ofpreceding orientations is surprising.

However, perceived test orientationmight be expected to depend notonly on the physical orientations ofthe adapting gratings, but also onthe perceived orientation of priorambiguous test stimuli [5,6]. Thus, itwas important that Chopin andMamassian [2] establish thegenerality of their findings beyondthe binocular rivalry paradigm. Theirsecond experiment made use of thetilt aftereffect, first reported 75 yearsago [7] but still a valuable tool tovision researchers, whereby theperceived orientation of a test gratingis biased by the diet of precedingorientations. Chopin and Mamassian[2] asked subjects to report to which ofthe two adapting orientations the testgrating was closer. Unbeknownst tothe subjects, the orientation of the testwas in fact always midway betweenthose of the adaptors. The resultsrevealed that subjects were less likelyto report the test orientation as beingcloser to the most recently presentedadaptors, but more likely to report itas closer to orientations seen2–10 minutes previously. The twoexperiments thus showed qualitativelysimilar patterns of results. In bothcases, the expected negativeaftereffect was observed at shortadaptor-test intervals but, for longerintervals of 5–10 minutes, the effectwas reversed and perceivedtest orientation correlated positively

with the orientations presented in thisearlier time window.This finding has important

implications for our understanding ofadaptation. Sensory adaptation hasbeen characterised as a process ofself-calibration of the system to itsenvironment [8–10]. Self-calibrationtheories of adaptation typically assumethat the visual system has someinternal model of the expecteddistribution of its response states.Deviations from this distributiondrive the system to modify thestimulus-response mapping. However,where this prior expectation of theresponse distribution comes from is farfrom clear. It is usually assumed toinvolve a longer timescale of learning,through the developmental stage of thelifespan or even evolving acrossgenerations. So to read evidence ofa timescale of 5–10 minutes issomewhat unexpected!Within a Bayesian framework [11,12],

the positive effects of adaptationsuggest that the visual system islearning the prior probabilities of thestimulus distribution over a longtimescale while the negative effectsrepresent a redistribution of sensoryresources in line with ideas of efficientcoding [13–15]. Thus, the history ofstimulation over this longer timescale istaken as predictive of future sensoryinput. Chopin and Mamassian [2] pointout that a similar strategy leads to thegambler’s fallacy in human reasoning[16]. Under the gambler’s fallacy, ifa coin comes up heads several times insuccession the ‘law of averages’suggests that it is more likely to comeup tails on the next toss. By matchingthe recent history of coin tosses to thereference distribution built up overa longer timescale (equal frequency ofheads and tails) the prediction of thegambler violates the expectation thatsuccessive coin tosses areindependent.In the context of sensory adaptation,

however, the same strategy couldprovide a means to correct forperturbations within the system. Forexample, if the sensory responseconsistently indicates more leftwardsthan rightwards oriented structure inthe environment, then this might betreated as an error within the systemthat can be corrected by turning downthe gain on the mechanism detectingleftwards orientation. This would causesubsequent stimuli to be more likelyto be perceived as oriented rightwards.

DispatchR225

If the recent history really hadconsisted predominantly of leftwardsorientation, then this would causea physically vertical stimulus to beperceived, fallaciously, as rightwards:the tilt aftereffect. However, undermore naturalistic viewing conditionssuch reliance on the statisticalstationarity of the structure of theenvironment might be an intelligentmeans to keep the sensory systemcalibrated in the face of internalfluctuations in excitability.

References1. Holland, J.H. (1992). Adaptation in Natural and

Artificial Systems (Cambridge, MA: MIT Press).2. Chopin, A., and Mamassian, P. (2012).

Predictive properties of visual adaptation.Curr. Biol. 22, 622–626.

3. Clifford, C.W.G. (2009). Binocular rivalry.Curr. Biol. 19, R1022–R1023.

4. Clifford, C.W.G. (2010). Aftereffects. InEncyclopedia of Perception, E.B. Goldstein, ed.

(Thousand Oaks, CA: Sage Publications Inc.),pp. 13–16.

5. Pearson, J., and Clifford, C.W.G. (2005).Mechanisms selectively engaged in rivalry:normal vision habituates, rivalrous visionprimes. Vis. Res. 45, 707–714.

6. Blake, R., Tadin, D., Sobel, K.V., Raissian, T.A.,and Chong, S.C. (2006). Strength of visualadaptation depends on visual awareness.Proc. Natl. Acad. Sci. USA 103,4783–4788.

7. Gibson, J.J., and Radner, M. (1937).Adaptation, after-effect, and contrast in theperception of tilted lines. I. Quantitativestudies. J. Exp. Psychol. 20, 453–467.

8. Andrews, D.P. (1964). Error-correctingperceptual mechanisms. Quart. J. Exp.Psychol. 16, 104–115.

9. Ullman, S., and Schechtman, G. (1982).Adaptation and gain normalization. Proc. R.Soc. Lond. B 216, 299–313.

10. Clifford, C.W.G. (2005). Functional ideas aboutadaptation applied to spatial and motion vision.In Fitting the Mind to the World: Aftereffects inHigh-Level Vision, C.W.G. Clifford andG. Rhodes, eds. (Oxford: Oxford UniversityPress), pp. 47–82.

11. Grzywacz, N.M., and Balboa, R.M. (2002). ABayesian framework for sensory adaptation.Neural Comput. 14, 543–559.

12. Stocker, A.A., and Simoncelli, E.P. (2006).Sensory adaptation within a Bayesianframework for perception. In Advances inNeural Information Processing Systems, Vol.18, Y. Weiss, B. Schoelkopf, and J. Platt, eds.(London: Springer), pp. 1291–1298.

13. Barlow, H.B. (1961). The coding of sensorymessages. In Current Problems in AnimalBehaviour, W.H. Thorpe and O.L. Zangwill, eds.(Cambridge: Cambridge University Press),pp. 331–360.

14. van Hateren, J.H. (1993). Spatiotemporalcontrast sensitivity of early vision. Vis. Res. 33,257–267.

15. Wainwright, M.J. (1999). Visual adaptation asoptimal information transmission. Vis. Res. 39,3960–3974.

16. Tversky, A., and Kahneman, D. (1974).Judgment under uncertainty: heuristics andbiases. Science 185, 1124–1131.

School of Psychology & Australian Centre ofExcellence in Vision Science, University ofSydney, Sydney, NSW 2006, Australia.E-mail: colin.clifford@usyd.edu.au

DOI: 10.1016/j.cub.2012.02.019

Fungal Morphogenesis: In Hot Pursuit

Temperature affects diverse biological processes. In fungi such as thepathogen Candida albicans, temperature governs a morphogenetic switchbetween yeast and hyphal growth. A new report connects the thermosensorHsp90 to a CDK–cyclin–transcription factor module that controlsmorphogenesis.

Wenjie Xu and Aaron P. Mitchell*

Morphogenesis — the development ofand transition between different growthforms — is a common theme in thefungal world. Morphogenetic pathwaysoften respond to environmental cues.For diverse pathogenic fungi, includingCandida albicans, Histoplasmacapsulatum, and Blastomycesdermatitidis, host body temperature(37�C) is a trigger of morphologicaltransitions. For C. albicans, the hyphalgrowth form that appears at hightemperature is also prominent ininfected tissue and is critical forC. albicans virulence [1]. Therefore,the thermal control over fungalmorphogenesis is an intriguing issuefrom the standpoint of both cell biologyand pathogenesis.

The molecular mechanismunderlying temperature control ofC. albicans morphogenesis remainedelusive until 2009. At that time, Cowenand colleagues [2] reported that Hsp90,a molecular chaperone with manyclient proteins, had a central role in

this mechanism. They found thatcompromising Hsp90 functionby genetic or pharmacologicalapproaches induced C. albicanscells to transit from the yeast formto the hyphal form, independentlyof temperature. It was particularlysignificant that Hsp90 was not simplyrequired to complete a developmentalprogram, as it is in many organisms,but rather that it seemed to governhyphal formation at the developmentaldecision point. These findings led tothe proposal that Hsp90 is a negativeregulator of hyphal morphogenesis,and that high temperature mayoverwhelm Hsp90 with client proteinsand thus relieve Hsp90-mediatedinhibition [2]. Therefore, Hsp90 itselffunctions as a temperaturesensor (Figure 1).

In a new study that recentlyappeared in Current Biology, Cowenand colleagues [3] implementednewly developed tools of C. albicansfunctional genomics to definea regulatory pathway that couplesthe Hsp90-dependent signal to

hyphal-specific gene activation. Theirprevious studies had implicatedthe cyclic AMP-protein kinase A(cAMP–PKA) pathway in this role [2].Surprisingly, though, the canonicaltranscription factor target ofcAMP–PKA, Efg1, was dispensablefor the response to Hsp90. Thus, thegroup set out to look specifically fortranscription factors that governHsp90-responsive morphogenesis.A library of 143 homozygous deletionmutants [4] was screened for a reducedcapacity to produce hyphae inresponse to the Hsp90 inhibitorgeldanamycin. The screen paid offbeautifully: a mutant lackinga previously uncharacterizedtranscription factor gene, HMS1,had a severe defect ingeldanamycin-induced hyphaeformation. Although the hyphalregulatory pathway is a complexmeshwork of interconnected signalsand transduction pathways [1], therole of Hms1 turns out to be quitespecific: it is required for induction ofhyphal formation by elevatedtemperature but not by otherenvironmental cues, such as nutritionallimitation, neutral pH, or serum.ChIP–chip and qRT–PCR analysesrevealed that Hms1 is bound to DNAassociated with hyphal-specific genes,such asUME6 andRBT5, and regulatestheir transcript levels. Together thesefinding indicate that Hms1 plays