- Email: [email protected]
Perceptual Learning: Complete Transfer across Retinal Locations
Current Biology Vol 18 No 24 R1134
can drive convergence in avian morphology [15,16]. The Loss of an Ancient Lineage The misleading taxonomy caused by this convergent evolution has been rectified by the new DNA-based analyses, which reveal the surprising uniqueness of these Hawaiian birds. The last sighting of a Hawaiian Chaetoptila occurred in 1859, and three of the four Moho species were likewise extinct by the mid-1900s. The final surviving species, Moho braccatus, persisted in the highlands of Kauai into the late 1980s, but is now almost certainly also extinct. The poignancy of this lineage’s decline is captured in recordings of the haunting song of the last (and mate-less) male Kauaii ‘O’o, which can be played online from the Cornell Lab of Ornithology’s sound archive (http://www.birds.cornell.edu/ macaulaylibrary/). Like so much of the native Hawaiian avifauna, these honeyeaters were doomed by a lethal combination of human-caused stressors [17]. The new phylogenetic evidence places the split between the Hawaiian honeyeaters and their living relatives at about 15 million years ago, a period coincident with the arrival in the islands of the bird-pollinated plants that likely fostered their nectivorous specializations [4,18]. The Hawaiian honeyeaters evolved their nectar-feeding adaptations and
spectacular plumages through this long period of evolutionary isolation. As suggested by Fleischer and his colleagues [4], the Hawaiian honeyeater lineage is best classified into its own new family, the Mohoidae. Sadly, this is the only avian family known to have gone extinct in its entirety in the past several centuries. Its demise therefore represents the loss of a particularly divergent evolutionary lineage [19,20], one that we only now recognize for its true level of uniqueness.
1. Futuyma, D.J. (2005). Evolution (Sunderland, MA: Sinauer Associates). 2. Harmon, L.J., Kolbe, J.J., Cheverud, J.M., and Losos, J.B. (2005). Convergence and the multidimensional niche. Evolution 59, 409–421. 3. Wake, D.B. (1991). Homoplasy: the result of natural selection or evidence of design limitations? Am. Nat. 138, 543–567. 4. Fleischer, R.C., James, H.F., and Olson, S.L. (2008). Convergent evolution of Hawaiian and Australo-Pacific honeyeaters from distant songbird ancestors. Curr. Biol. 18, 1927–1931. 5. James, H.F., and Olson, S.L. (1991). Descriptions of thirty-two new species of birds from the Hawaiian Islands: Part II. Passerines. Ornithological Monographs 46, 1–88. 6. Fleischer, R.C., and McIntosh, C.E. (2001). Molecular systematics and biogeography of the Hawaiian avifauna. Studies Avian Biol. 22, 51–60. 7. Fuller, E. (2001). Extinct Birds (Ithaca, NY: Cornell University Press). 8. Dickinson, E.C. (2003). The Howard and Moore Complete Checklist of the Birds of the World, 3rd edition (London: Christopher Helm). 9. Barker, F.K., Cibois, A., Schikler, P., Feinstein, J., and Cracraft, J. (2004). Phylogeny and diversification of the largest avian
A newly developed ‘double training’ technique demonstrates that practicedependent improvement in the discrimination of basic visual features will transfer to a location that has been trained with a different discrimination.
Gibson [1] defined perceptual learning as ‘‘any relatively permanent and consistent change in the perception of a stimulus array, following practice or experience with this array’’. This definition encompasses practice-dependent improvements in performance ranging from the observation that experienced wine-tasters can make
11.
12.
13. 14.
References
Perceptual Learning: Complete Transfer across Retinal Locations
Dominic M. Dwyer
10.
distinctions between complex multimodal stimuli that are not possible for novices [2] to the fact that extensive practice with very simple discriminations, for example between the Vernier offsets of two sets of lines, leads to improved performance [3]. There is much evidence that perceptual learning with simple stimuli can be very specific to the training situation: changes in the retinal
15. 16.
17.
18.
19. 20.
radiation. Proc. Natl. Acad. Sci. USA 101, 11040–11045. Hackett, S.J., Kimball, R.T., Reddy, S., Bowie, R.C., Braun, E.L., Braun, M.J., Chojnowski, J.L., Cox, W.A., Han, K.L, Harshman, J., et al. (2008). A phylogenomic study of birds reveals their evolutionary history. Science 320, 1763–1768. Fuchs, J., Fjeldsa, J., Bowie, R.C.K., Voelker, G., and Pasquet, E. (2006). The African warbler genus Hyliota as a lost lineage in the Oscine songbird tree: Molecular support for an African origin of the Passerida. Mol. Phyl. Evol. 39, 186–197. Cibois, A., and Cracraft, J. (2004). Assessing the passerine ‘‘Tapestry’’: phylogenetic relationships of the Muscicapoidea inferred from nuclear DNA sequences. Mol. Phyl. Evol. 32, 264–273. Willerslev, E., and Cooper, A. (2005). Ancient DNA. Proc. Royal. Soc. Lond. B 272, 3–16. Millar, C.D., Huynen, L., Subramanian, S., Mohandesan, E., David, M., and Lambert, D.M. (2008). New developments in ancient genomics. Trends Ecol. Evol. 23, 386–393. Beecher, W.J. (1951). Convergence in the Coerebidae. Wilson Bull. 63, 274–287. Burns, K.J., Hackett, S.J., and Klein, N.K. (2003). Phylogenetic relationships of Neotropical honeycreepers and the evolution of feeding morphology. J. Avian Biol. 34, 360–370. Boyer, A.G. (2008). Extinction patterns in the avifauna of the Hawaiian Islands. Divers. Distrib. 14, 509–517. Price, J.P., and Clague, D.A. (2002). How old is the Hawaiian biota? Geology and phylogeny suggest recent divergence. Proc. Roy. Soc. B 269, 2429–2435. Faith, D.P. (1992). Conservation evaluation and phylogenetic diversity. Biol. Cons. 61, 1–10. Erwin, D.H. (2008). Extinction as the loss of evolutionary history. Proc. Natl. Acad. Sci. USA 105, 11520–11527.
Fuller Evolutionary Biology Program, Cornell Lab of Ornithology, Cornell University, 159 Sapsucker Woods Road, Ithaca, NY 14850, USA. E-mail: [email protected]
DOI: 10.1016/j.cub.2008.11.006
location or orientation of the lines between training and test greatly reduces or abolishes the effect of practicing Vernier discriminations or other simple discrimination tasks [3–5]. Such specificity is unlikely to be present with more complex stimuli, such as faces [6,7], that are not restricted to a single retinal location. Research reported by Xiao et al. [8] in this issue of Current Biology questions whether perceptual learning with simple visual stimuli is genuinely specific to particular retinal locations by demonstrating that, with appropriate training methods, improvements in discrimination can transfer completely across locations. The issue of whether or not perceptual learning is specific to trained locations or stimuli is of
Dispatch R1135
+
E
D
C
B
A
+
+
+
Test after phase 1 training
Test after phase 2 training
40 30 Mean % improvement
critical importance because early visual cortex contains neurons that are more selective to the position and orientation of stimuli than are neurons further down the visual processing stream. Thus, the apparent specificity of perceptual learning to particular orientations or locations was considered to be very strong evidence that the neural mechanisms underpinning the improvements must involve the early visual cortex (for example [9,10]). So consistent have the observations of stimulus specificity been that they have, quite justifiably, been taken as a grounding constraint on both empirical investigations and theoretical accounts of perceptual learning (for example [10,11]). Despite this, it was noted over a decade ago that the specificity in perceptual learning might lie in what is learnt, rather than where the learning takes place [12], and this possibility has only now been directly investigated [8]. A simplified version of the double-training version of the Vernier discrimination task used by Xiao et al. [8] is shown in Figure 1. Initially training was given with lines of one orientation in one location (for example, vertical lines in the upper left visual quadrant). In line with previous results, this produced an improvement in discrimination performance, but this improvement did not transfer to horizontal lines tested at the original location or to the vertical lines tested at a different location such as the lower left visual quadrant (see the leftmost three bars in Figure 1E). Subsequently, training was given with a second orientation at a second location — for example, horizontal lines in the lower left visual quadrant — before testing with the assignment of line orientation to training location reversed. This second phase of training dramatically changed the results of the critical transfer tests. Discrimination with the vertical lines in the lower quadrant was now as good as it was in the upper quadrant at which it had actually been trained. Discrimination with horizontal lines was as good in the upper quadrant, where it had not been trained, as it was in the lower quadrant where it had been trained (see the rightmost three bars Figure 1E). Other experiments confirmed that double-training allowed for complete transfer of perceptual learning across
20 10 0 –10 –20 Ori1 Loc1
Ori1 Loc2
Ori2 Loc1
Ori2 Loc2
Ori1 Loc2
Ori2 Loc1 Current Biology
Figure 1. A simplified version of Experiment 3 from [8]. (A) Phase 1 training with Vernier discriminations using vertical lines in the upper left quadrant; orientation 1-location 1 (Ori1Loc1 in (E), which shows the percent improvement over baseline in this condition). (B) The transfer test condition orientation 1-location 2 (Ori1Loc2): Vernier discrimination with the original training stimulus at a new location. (C) Transfer test condition orientation 2-location 1 (Ori2Loc1): Vernier discrimination with a new stimulus at the original location. (D) Phase 2 training with Vernier discriminations using horizontal lines in the lower left quadrant (orientation 2-location 2, Ori2Loc2). The transfer tests shown in (B) and (C) were performed after both phase 1 and phase 2 training (left and right sides of (E), respectively). Note, in the actual experiment the locations and line orientations were counterbalanced. (E) Summary results as mean percentage improvement in discrimination performance over baseline. The left side shows performance after phase 1 training and the right side shows performance after phase 2 training.
locations when the two discriminations were trained concurrently and when different types of discrimination, rather than different orientations of lines, were used. The complete transfer of training-dependent improvement in discrimination from one retinal location to another directly challenges the idea that location specificity is a key feature of perceptual learning. In turn, this questions the common belief that retinotopically organised early visual cortex is the neural site for perceptual learning and suggests instead that more central mechanisms are involved. This is not to say that perceptual learning does not influence primary visual cortex, as there is direct evidence that it does (for example
[13,14]), but that the critical site for the process rather than the outcome of perceptual learning must be more central or complete transfer across locations would be impossible. Location transfer also implies that a reconsideration is needed of the role played by non-retinotipic central brain mechanisms (for example [15]). These have previously been considered in light of the fact that attention constrains perceptual learning (for example [16]), but the results of Xiao et al. [8] suggest that they might actually underpin the learning itself rather than simply its attentional modulation. Retinotopic specificity would constrain not only neural models of
Current Biology Vol 18 No 24 R1136
perceptual learning but also functional ones. A number of theorists have argued that perceptual learning should be considered as independent of more general processes which would not be restricted to stimulus-specific features or locations (for example [10,17]). Relaxing this constraint gives additional support to the alternative view that more general mechanisms such as association formation or categorization can make significant contributions to perceptual learning (for example [18–20]). Similarly, by demonstrating that learning with simple stimuli can be independent of location in the same way as more complex stimuli, Xiao et al.’s [8] work raises the possibility that perceptual learning with simple and complex stimuli might rely on at least partially overlapping mechanisms. In summary, by implicating central, rather than peripheral, mechanisms for perceptual learning with simple visual stimuli the demonstration of complete transfer across retinal locations raises many interesting possibilities. In particular, that there might be more commonalities between perceptual learning with simple and complex stimuli and between general cognitive mechanisms and their perceptual
consequences than have previously been supposed. References 1. Gibson, E.J. (1963). Perceptual learning. Annu. Rev. Psychol. 14, 29–56. 2. Solomon, G.E.A. (1997). Conceptual change and wine expertise. J. Learn. Sci. 6, 41–60. 3. Fahle, M. (1997). Specificity of learning curvature, orientation, and vernier discriminations. Vision Res. 37, 1885–1895. 4. Shiu, L.P., and Pashler, H. (1992). Improvement in line orientation discrimination is retinally local but dependent on cognitive set. Percept. Psychophys. 52, 582–588. 5. Karni, A., and Sagi, D. (1991). Where practice makes perfect in texture-discrimination evidence for primary visual-cortex plasticity. Proc. Natl. Acad. Sci. USA 88, 4966–4970. 6. Mundy, M.E., Honey, R.C., and Dwyer, D.M. (2007). Simultaneous presentation of similar stimuli produces perceptual learning in human picture processing. J. Exp. Psychol. 33, 124–138. 7. Bruce, V., and Burton, A.M. (2002). Learning new faces. In Perceptual Learning, T. Poggio and M. Fahle, eds. (Cambridge, MA: MIT Press), pp. 317–334. 8. Xiao, L., Zhang, J., Wang, R., Klein, S.A., Levi, D.M., and Yu, C. (2008). Complete transfer of perceptual learning across retinal locations enables by double training. Curr. Biol. 18, 1922–1926. 9. Fahle, M. (2004). Perceptual learning: A case for early selection. J. Vision 4, 879–890. 10. Fahle, M. (2002). Introduction. In Perceptual Learning, M. Fahle and T. Poggio, eds. (Cambridge, MA: MIT Press), pp. ix–xx. 11. Dosher, B.A., and Lu, Z.L. (1999). Mechanisms of perceptual learning. Vision Res. 39, 3197–3221. 12. Mollon, J.D., and Danilova, M.V. (1996). Three remarks on perceptual learning. Spatial Vision 10, 51–58.
Gene Expression: Dialing Up the Frequency Cells often respond to external signals by altering their gene expression. The external signaling information is transduced and typically encoded in concentrations of relevant transcription factors. A recent study demonstrates that, by encoding this information in the frequency with which genes ‘see’ a transcription factor, the expression of hundreds of genes can be modulated in a linearly proportional manner. Narendra Maheshri The single input module is a prevalent network motif in genetic regulatory networks that allows cells to respond to external signals through the coordinated regulation of hundreds of genes. This module consists of a transcription factor (TF) that directly regulates the expression of many downstream genes. Typically, external signal information is encoded in the concentration of the TF. Each downstream gene responds to TF
levels in a different way, depending on the details of the promoter. A gene regulatory function is a compact mathematical way to represent the response of each gene to different TF concentrations [1]. These responses are typically hyperbolic or sigmoidal and can be described by a Hill-like n , where function: w k ½TF½TF n + Kn k corresponds to the strength of the promoter, K is the affinity of TF–promoter binding, and the Hill coefficient n captures the degree of cooperativity in TF–promoter binding.
13. Schoups, A., Vogels, R., Qian, N., and Orban, G. (2001). Practising orientation identification improves orientation coding in V1 neurons. Nature 412, 549–553. 14. Schwartz, S., Maquet, P., and Frith, C. (2002). Neural correlates of perceptual learning: A functional MIR study of visual texture discrimination. Proc. Natl. Acad. Sci. USA 99, 17137–17142. 15. Mukai, I., Kim, D., Fukunaga, M., Japee, S., Marrett, S., and Ungerleider, L.G. (2007). Activations in visual and attention-related areas predict and correlate with the degree of perceptual learning. J. Neurosci. 27, 11401–11411. 16. Ahissar, M., and Hochstein, S. (1993). Attentional control of early perceptual-learning. Proc. Natl. Acad. Sci. USA 90, 5718–5722. 17. Hall, G. (1991). Perceptual and Associative Learning (Oxford: Oxford University Press). 18. McLaren, I.P.L., and Mackintosh, N.J. (2000). An elemental model of associative learning: I. Latent inhibition and perceptual learning. Anim. Learn. Behav. 28, 211–246. 19. Goldstone, R. (2003). Learning to perceive while perceiving to learn. In Perceptual Organization in Vision: Behavioral and Neural Perspectives, R. Kimchi, M. Behrmann, and C. Olson, eds. (Mahwah, NJ: Lawrence Erlbaum Associates), pp. 233–280. 20. Mundy, M.E., Dwyer, D.M., and Honey, R.C. (2006). Inhibitory associations contribute to perceptual learning in humans. J. Exp. Psychol. 32, 178–184.
School of Psychology, Cardiff University, Tower Building, Park Place, Cardiff CF10 3AT, UK. E-mail: [email protected]
DOI: 10.1016/j.cub.2008.10.037
Differential expression is then due to the various affinities of each promoter within the single input module. By encoding signal information within TF concentrations and response information within promoters, cells are capable of executing regulatory programs that coordinate the timing of expression of hundreds of genes. For example, if the TF within a single input module is autoregulated by itself or its targets, the external signal triggers a slow rise of the TF, which turns on high-affinity (low K) genes early and low-affinity (high K) genes late. Some examples of this strategy include precise timing in developmental systems [2], flagellar biosynthesis in Escherichia coli [3], and host and viral gene expression post-infection [4]. However, what if the goal is to double the expression of all downstream genes in response to a change in an external signal?
can drive convergence in avian morphology [15,16]. The Loss of an Ancient Lineage The misleading taxonomy caused by this convergent evolution has been rectified by the new DNA-based analyses, which reveal the surprising uniqueness of these Hawaiian birds. The last sighting of a Hawaiian Chaetoptila occurred in 1859, and three of the four Moho species were likewise extinct by the mid-1900s. The final surviving species, Moho braccatus, persisted in the highlands of Kauai into the late 1980s, but is now almost certainly also extinct. The poignancy of this lineage’s decline is captured in recordings of the haunting song of the last (and mate-less) male Kauaii ‘O’o, which can be played online from the Cornell Lab of Ornithology’s sound archive (http://www.birds.cornell.edu/ macaulaylibrary/). Like so much of the native Hawaiian avifauna, these honeyeaters were doomed by a lethal combination of human-caused stressors [17]. The new phylogenetic evidence places the split between the Hawaiian honeyeaters and their living relatives at about 15 million years ago, a period coincident with the arrival in the islands of the bird-pollinated plants that likely fostered their nectivorous specializations [4,18]. The Hawaiian honeyeaters evolved their nectar-feeding adaptations and
spectacular plumages through this long period of evolutionary isolation. As suggested by Fleischer and his colleagues [4], the Hawaiian honeyeater lineage is best classified into its own new family, the Mohoidae. Sadly, this is the only avian family known to have gone extinct in its entirety in the past several centuries. Its demise therefore represents the loss of a particularly divergent evolutionary lineage [19,20], one that we only now recognize for its true level of uniqueness.
1. Futuyma, D.J. (2005). Evolution (Sunderland, MA: Sinauer Associates). 2. Harmon, L.J., Kolbe, J.J., Cheverud, J.M., and Losos, J.B. (2005). Convergence and the multidimensional niche. Evolution 59, 409–421. 3. Wake, D.B. (1991). Homoplasy: the result of natural selection or evidence of design limitations? Am. Nat. 138, 543–567. 4. Fleischer, R.C., James, H.F., and Olson, S.L. (2008). Convergent evolution of Hawaiian and Australo-Pacific honeyeaters from distant songbird ancestors. Curr. Biol. 18, 1927–1931. 5. James, H.F., and Olson, S.L. (1991). Descriptions of thirty-two new species of birds from the Hawaiian Islands: Part II. Passerines. Ornithological Monographs 46, 1–88. 6. Fleischer, R.C., and McIntosh, C.E. (2001). Molecular systematics and biogeography of the Hawaiian avifauna. Studies Avian Biol. 22, 51–60. 7. Fuller, E. (2001). Extinct Birds (Ithaca, NY: Cornell University Press). 8. Dickinson, E.C. (2003). The Howard and Moore Complete Checklist of the Birds of the World, 3rd edition (London: Christopher Helm). 9. Barker, F.K., Cibois, A., Schikler, P., Feinstein, J., and Cracraft, J. (2004). Phylogeny and diversification of the largest avian
A newly developed ‘double training’ technique demonstrates that practicedependent improvement in the discrimination of basic visual features will transfer to a location that has been trained with a different discrimination.
Gibson [1] defined perceptual learning as ‘‘any relatively permanent and consistent change in the perception of a stimulus array, following practice or experience with this array’’. This definition encompasses practice-dependent improvements in performance ranging from the observation that experienced wine-tasters can make
11.
12.
13. 14.
References
Perceptual Learning: Complete Transfer across Retinal Locations
Dominic M. Dwyer
10.
distinctions between complex multimodal stimuli that are not possible for novices [2] to the fact that extensive practice with very simple discriminations, for example between the Vernier offsets of two sets of lines, leads to improved performance [3]. There is much evidence that perceptual learning with simple stimuli can be very specific to the training situation: changes in the retinal
15. 16.
17.
18.
19. 20.
radiation. Proc. Natl. Acad. Sci. USA 101, 11040–11045. Hackett, S.J., Kimball, R.T., Reddy, S., Bowie, R.C., Braun, E.L., Braun, M.J., Chojnowski, J.L., Cox, W.A., Han, K.L, Harshman, J., et al. (2008). A phylogenomic study of birds reveals their evolutionary history. Science 320, 1763–1768. Fuchs, J., Fjeldsa, J., Bowie, R.C.K., Voelker, G., and Pasquet, E. (2006). The African warbler genus Hyliota as a lost lineage in the Oscine songbird tree: Molecular support for an African origin of the Passerida. Mol. Phyl. Evol. 39, 186–197. Cibois, A., and Cracraft, J. (2004). Assessing the passerine ‘‘Tapestry’’: phylogenetic relationships of the Muscicapoidea inferred from nuclear DNA sequences. Mol. Phyl. Evol. 32, 264–273. Willerslev, E., and Cooper, A. (2005). Ancient DNA. Proc. Royal. Soc. Lond. B 272, 3–16. Millar, C.D., Huynen, L., Subramanian, S., Mohandesan, E., David, M., and Lambert, D.M. (2008). New developments in ancient genomics. Trends Ecol. Evol. 23, 386–393. Beecher, W.J. (1951). Convergence in the Coerebidae. Wilson Bull. 63, 274–287. Burns, K.J., Hackett, S.J., and Klein, N.K. (2003). Phylogenetic relationships of Neotropical honeycreepers and the evolution of feeding morphology. J. Avian Biol. 34, 360–370. Boyer, A.G. (2008). Extinction patterns in the avifauna of the Hawaiian Islands. Divers. Distrib. 14, 509–517. Price, J.P., and Clague, D.A. (2002). How old is the Hawaiian biota? Geology and phylogeny suggest recent divergence. Proc. Roy. Soc. B 269, 2429–2435. Faith, D.P. (1992). Conservation evaluation and phylogenetic diversity. Biol. Cons. 61, 1–10. Erwin, D.H. (2008). Extinction as the loss of evolutionary history. Proc. Natl. Acad. Sci. USA 105, 11520–11527.
Fuller Evolutionary Biology Program, Cornell Lab of Ornithology, Cornell University, 159 Sapsucker Woods Road, Ithaca, NY 14850, USA. E-mail: [email protected]
DOI: 10.1016/j.cub.2008.11.006
location or orientation of the lines between training and test greatly reduces or abolishes the effect of practicing Vernier discriminations or other simple discrimination tasks [3–5]. Such specificity is unlikely to be present with more complex stimuli, such as faces [6,7], that are not restricted to a single retinal location. Research reported by Xiao et al. [8] in this issue of Current Biology questions whether perceptual learning with simple visual stimuli is genuinely specific to particular retinal locations by demonstrating that, with appropriate training methods, improvements in discrimination can transfer completely across locations. The issue of whether or not perceptual learning is specific to trained locations or stimuli is of
Dispatch R1135
+
E
D
C
B
A
+
+
+
Test after phase 1 training
Test after phase 2 training
40 30 Mean % improvement
critical importance because early visual cortex contains neurons that are more selective to the position and orientation of stimuli than are neurons further down the visual processing stream. Thus, the apparent specificity of perceptual learning to particular orientations or locations was considered to be very strong evidence that the neural mechanisms underpinning the improvements must involve the early visual cortex (for example [9,10]). So consistent have the observations of stimulus specificity been that they have, quite justifiably, been taken as a grounding constraint on both empirical investigations and theoretical accounts of perceptual learning (for example [10,11]). Despite this, it was noted over a decade ago that the specificity in perceptual learning might lie in what is learnt, rather than where the learning takes place [12], and this possibility has only now been directly investigated [8]. A simplified version of the double-training version of the Vernier discrimination task used by Xiao et al. [8] is shown in Figure 1. Initially training was given with lines of one orientation in one location (for example, vertical lines in the upper left visual quadrant). In line with previous results, this produced an improvement in discrimination performance, but this improvement did not transfer to horizontal lines tested at the original location or to the vertical lines tested at a different location such as the lower left visual quadrant (see the leftmost three bars in Figure 1E). Subsequently, training was given with a second orientation at a second location — for example, horizontal lines in the lower left visual quadrant — before testing with the assignment of line orientation to training location reversed. This second phase of training dramatically changed the results of the critical transfer tests. Discrimination with the vertical lines in the lower quadrant was now as good as it was in the upper quadrant at which it had actually been trained. Discrimination with horizontal lines was as good in the upper quadrant, where it had not been trained, as it was in the lower quadrant where it had been trained (see the rightmost three bars Figure 1E). Other experiments confirmed that double-training allowed for complete transfer of perceptual learning across
20 10 0 –10 –20 Ori1 Loc1
Ori1 Loc2
Ori2 Loc1
Ori2 Loc2
Ori1 Loc2
Ori2 Loc1 Current Biology
Figure 1. A simplified version of Experiment 3 from [8]. (A) Phase 1 training with Vernier discriminations using vertical lines in the upper left quadrant; orientation 1-location 1 (Ori1Loc1 in (E), which shows the percent improvement over baseline in this condition). (B) The transfer test condition orientation 1-location 2 (Ori1Loc2): Vernier discrimination with the original training stimulus at a new location. (C) Transfer test condition orientation 2-location 1 (Ori2Loc1): Vernier discrimination with a new stimulus at the original location. (D) Phase 2 training with Vernier discriminations using horizontal lines in the lower left quadrant (orientation 2-location 2, Ori2Loc2). The transfer tests shown in (B) and (C) were performed after both phase 1 and phase 2 training (left and right sides of (E), respectively). Note, in the actual experiment the locations and line orientations were counterbalanced. (E) Summary results as mean percentage improvement in discrimination performance over baseline. The left side shows performance after phase 1 training and the right side shows performance after phase 2 training.
locations when the two discriminations were trained concurrently and when different types of discrimination, rather than different orientations of lines, were used. The complete transfer of training-dependent improvement in discrimination from one retinal location to another directly challenges the idea that location specificity is a key feature of perceptual learning. In turn, this questions the common belief that retinotopically organised early visual cortex is the neural site for perceptual learning and suggests instead that more central mechanisms are involved. This is not to say that perceptual learning does not influence primary visual cortex, as there is direct evidence that it does (for example
[13,14]), but that the critical site for the process rather than the outcome of perceptual learning must be more central or complete transfer across locations would be impossible. Location transfer also implies that a reconsideration is needed of the role played by non-retinotipic central brain mechanisms (for example [15]). These have previously been considered in light of the fact that attention constrains perceptual learning (for example [16]), but the results of Xiao et al. [8] suggest that they might actually underpin the learning itself rather than simply its attentional modulation. Retinotopic specificity would constrain not only neural models of
Current Biology Vol 18 No 24 R1136
perceptual learning but also functional ones. A number of theorists have argued that perceptual learning should be considered as independent of more general processes which would not be restricted to stimulus-specific features or locations (for example [10,17]). Relaxing this constraint gives additional support to the alternative view that more general mechanisms such as association formation or categorization can make significant contributions to perceptual learning (for example [18–20]). Similarly, by demonstrating that learning with simple stimuli can be independent of location in the same way as more complex stimuli, Xiao et al.’s [8] work raises the possibility that perceptual learning with simple and complex stimuli might rely on at least partially overlapping mechanisms. In summary, by implicating central, rather than peripheral, mechanisms for perceptual learning with simple visual stimuli the demonstration of complete transfer across retinal locations raises many interesting possibilities. In particular, that there might be more commonalities between perceptual learning with simple and complex stimuli and between general cognitive mechanisms and their perceptual
consequences than have previously been supposed. References 1. Gibson, E.J. (1963). Perceptual learning. Annu. Rev. Psychol. 14, 29–56. 2. Solomon, G.E.A. (1997). Conceptual change and wine expertise. J. Learn. Sci. 6, 41–60. 3. Fahle, M. (1997). Specificity of learning curvature, orientation, and vernier discriminations. Vision Res. 37, 1885–1895. 4. Shiu, L.P., and Pashler, H. (1992). Improvement in line orientation discrimination is retinally local but dependent on cognitive set. Percept. Psychophys. 52, 582–588. 5. Karni, A., and Sagi, D. (1991). Where practice makes perfect in texture-discrimination evidence for primary visual-cortex plasticity. Proc. Natl. Acad. Sci. USA 88, 4966–4970. 6. Mundy, M.E., Honey, R.C., and Dwyer, D.M. (2007). Simultaneous presentation of similar stimuli produces perceptual learning in human picture processing. J. Exp. Psychol. 33, 124–138. 7. Bruce, V., and Burton, A.M. (2002). Learning new faces. In Perceptual Learning, T. Poggio and M. Fahle, eds. (Cambridge, MA: MIT Press), pp. 317–334. 8. Xiao, L., Zhang, J., Wang, R., Klein, S.A., Levi, D.M., and Yu, C. (2008). Complete transfer of perceptual learning across retinal locations enables by double training. Curr. Biol. 18, 1922–1926. 9. Fahle, M. (2004). Perceptual learning: A case for early selection. J. Vision 4, 879–890. 10. Fahle, M. (2002). Introduction. In Perceptual Learning, M. Fahle and T. Poggio, eds. (Cambridge, MA: MIT Press), pp. ix–xx. 11. Dosher, B.A., and Lu, Z.L. (1999). Mechanisms of perceptual learning. Vision Res. 39, 3197–3221. 12. Mollon, J.D., and Danilova, M.V. (1996). Three remarks on perceptual learning. Spatial Vision 10, 51–58.
Gene Expression: Dialing Up the Frequency Cells often respond to external signals by altering their gene expression. The external signaling information is transduced and typically encoded in concentrations of relevant transcription factors. A recent study demonstrates that, by encoding this information in the frequency with which genes ‘see’ a transcription factor, the expression of hundreds of genes can be modulated in a linearly proportional manner. Narendra Maheshri The single input module is a prevalent network motif in genetic regulatory networks that allows cells to respond to external signals through the coordinated regulation of hundreds of genes. This module consists of a transcription factor (TF) that directly regulates the expression of many downstream genes. Typically, external signal information is encoded in the concentration of the TF. Each downstream gene responds to TF
levels in a different way, depending on the details of the promoter. A gene regulatory function is a compact mathematical way to represent the response of each gene to different TF concentrations [1]. These responses are typically hyperbolic or sigmoidal and can be described by a Hill-like n , where function: w k ½TF½TF n + Kn k corresponds to the strength of the promoter, K is the affinity of TF–promoter binding, and the Hill coefficient n captures the degree of cooperativity in TF–promoter binding.
13. Schoups, A., Vogels, R., Qian, N., and Orban, G. (2001). Practising orientation identification improves orientation coding in V1 neurons. Nature 412, 549–553. 14. Schwartz, S., Maquet, P., and Frith, C. (2002). Neural correlates of perceptual learning: A functional MIR study of visual texture discrimination. Proc. Natl. Acad. Sci. USA 99, 17137–17142. 15. Mukai, I., Kim, D., Fukunaga, M., Japee, S., Marrett, S., and Ungerleider, L.G. (2007). Activations in visual and attention-related areas predict and correlate with the degree of perceptual learning. J. Neurosci. 27, 11401–11411. 16. Ahissar, M., and Hochstein, S. (1993). Attentional control of early perceptual-learning. Proc. Natl. Acad. Sci. USA 90, 5718–5722. 17. Hall, G. (1991). Perceptual and Associative Learning (Oxford: Oxford University Press). 18. McLaren, I.P.L., and Mackintosh, N.J. (2000). An elemental model of associative learning: I. Latent inhibition and perceptual learning. Anim. Learn. Behav. 28, 211–246. 19. Goldstone, R. (2003). Learning to perceive while perceiving to learn. In Perceptual Organization in Vision: Behavioral and Neural Perspectives, R. Kimchi, M. Behrmann, and C. Olson, eds. (Mahwah, NJ: Lawrence Erlbaum Associates), pp. 233–280. 20. Mundy, M.E., Dwyer, D.M., and Honey, R.C. (2006). Inhibitory associations contribute to perceptual learning in humans. J. Exp. Psychol. 32, 178–184.
School of Psychology, Cardiff University, Tower Building, Park Place, Cardiff CF10 3AT, UK. E-mail: [email protected]
DOI: 10.1016/j.cub.2008.10.037
Differential expression is then due to the various affinities of each promoter within the single input module. By encoding signal information within TF concentrations and response information within promoters, cells are capable of executing regulatory programs that coordinate the timing of expression of hundreds of genes. For example, if the TF within a single input module is autoregulated by itself or its targets, the external signal triggers a slow rise of the TF, which turns on high-affinity (low K) genes early and low-affinity (high K) genes late. Some examples of this strategy include precise timing in developmental systems [2], flagellar biosynthesis in Escherichia coli [3], and host and viral gene expression post-infection [4]. However, what if the goal is to double the expression of all downstream genes in response to a change in an external signal?