Pelli Lab

Selected Conference Presentations (Posters and Talks)

2014 | 2013 | 2012 | 2011 | 2010 | 2009 | 2008 | 2007 | 2006 | 2005 | 2004 | 2003 | 2002 | 2001 | 2000 | 1999 | 1998 | 1997 | 1996 | 1995 | 1994 | 1993 | 1992 | 1991 | 1990 | 1989 | 1988 | 1987 | 1986 | 1985 | 1984 | 1983 | 1982 | 1981

poster thumbnail Vale, L., and Pelli, D. G. (2014) The duration of pleasure in the experience of beauty. Vision Sciences Society, St. Pete Beach, Florida, May 16-21, 2014, 63.328. pdf
poster thumbnail Ziskind, A.J., Hénaff, O., LeCun, Y., & Pelli, D.G. (2014) The bottleneck in human letter recognition: A computational model. Vision Sciences Society, St. Pete Beach, Florida, May 16-21, 2014, 56.583. pdf
poster thumbnail A.J. Ziskind, Y. LeCun, D.G. Pelli (2014) Two machine-learning models of object recognition exhibit key features of human performance. Moore-Sloan Data Science Initiative Launch Event, NYU, New York, April 28, 2014. http://cds.nyu.edu/projects/two-machine-learning-models-object-recognition-exhibit-key-features-human-performance/ pdf
S. Rosen & D.G. Pelli (2012) Reading faster by reducing visual crowding. Vision Sciences Society Annual Meeting 2012, May 10-15, 2012, 34.12
http://f1000.com/posters/browse/summary/1090326
 
poster thumbnail D. G. Pelli, S. Song, D. M. Levi (2011) Improving the screening of children for amblyopia. Vision Sciences Society, Naples, Florida, May 6-11, 2011. pdf
poster Jacob, M., Rosen, S., & Pelli, D. G. (2011) The way we see it: How familiarity affects perception. NYU Undergraduate Research Conference. New York City, April 15, 2011. pdf
poster Schnebelen, W., Tillman, K. A., Dubois, M., & Pelli, D. G. (2011) As children learn to read, eye and ear improve but always integrate well. NYU Undergraduate Research Conference. New York City, April 15, 2011. pdf
Pelli, D. G., Freeman, J., and Chakravarthi, R. (2010). Crowding combines. Vision Sciences Society. Naples, Florida. May 7-12, 2010.

ABSTRACT. Visual crowding provides a window into object recognition: observers fail to recognize objects in clutter. Here we ask, what do they see instead? We analyze observers’ errors to show that crowding necessarily reflects the combination of information across multiple complex objects, rather than the mislocalization (or substitution) of one object for another. First, we presented single letters, randomly chosen, in noise in the periphery and tabulated a confusion matrix based on observers’ (n=3) reports. We then tested the same observers in a classic crowding task, in which they viewed a triplet (target and two flankers) of closely spaced letters in the periphery (10 deg) and reported the identity of the middle target. For each observer, we tailored the triplets based on that observer’s single-letter confusion matrix. One flanker was chosen to be a letter that was most confused with (most “similar” to) to the target, and the other was chosen to be a letter that was least confused (least similar). Consistent with the literature, when mistaken, observers tend to report the flankers. The crucial issue, however, is which of the two flankers observers report on these trials. Blind substitu- tion predicts that the two flankers (similar and dissimilar) are equally likely to be reported. Instead, we find that observers are more likely to report the similar flanker (70%) than the dissimilar flanker (30%). The effect of similar- ity on erroneous responses proves that the response combines information from both the target and the reported flanker. By systematically tailoring the stimuli, we induced a bias in the reports that reveals a pooled, “mon- grel-like,” underlying percept. Our method, applicable to any object, gen- eralizes the evidence for “compulsory pooling” from the narrow domain of grating orientation (Parkes et al., 2001) to complex, everyday objects.

pdf

View the one-hour video at:

http:futureofreading.cias.rit.edu/2010/video.php@DennisPelli

One-hour talk "The role of vision in reading" presented at the "Future of Reading 2010 Symposium" at Rochester Institute of Technology. Program of Symposium

The Role of Vision in Reading
We’ve all been read to as children. And blind people learn to read braille. But, mostly, reading is a visual act. The advance of printing and display technology continues unabated, with large social consequences, especially near-universal literacy. And now, as Cody Brown says, publishing is the new literacy. Still, we all read through human eyes, by optically imaging text onto our retinas. We read by recognizing a serial stream of words. Each word is an object. The limitations of object recognition limit reading. The two most important limitations are letter size and spacing (center to center). Those two factors limit reading speed and describe most of the variation in reading by normal and clinical populations. Visual object recognition will constrain reading for the foreseeable future.

 
poster Rosen, S., Chakravarthi, R., and Pelli, D. G. (2010). Pool party: Objects rule! Vision Sciences Society. Naples, Florida. May 7-12, 2010.

 

pdf
poster Granata, Y., Chakravarthi, R., Rosen, S., and Pelli, D. G. (2010). Size pooling. Vision Sciences Society. Naples, Florida. May 7-12, 2010.

 

pdf
poster Raghavan, M., Remler, B. F., Rozman, S., & Pelli, D. G. (2010). Patients with visual "snow" have normal equivalent input noise levels. Investigative Ophthalmology & Visual Science, 51, ARVO E-Abstract 1808/D1660.

ABSTRACT. Purpose: "Visual snow" is a poorly understood symptom where patients report seeing fine-grained flickering spots as a chronic aspect of their visual experience. The symptom may be acquired, self-remitting, or lifelong. We hypothesize that what the patients see as "snow" is their own intrinsic visual noise. Our experiments assess whether visual-snow patients have increased levels of intrinsic noise. Methods: We quantified intrinsic visual noise in 5 patients with visual snow as an equivalent input noise at the display by measuring grating detection thresholds, with and without high levels of added dynamic display noise (Pelli and Farell, 1999). In central vision, retinal photon noise and cortical neural noise dominate in different signal domains. With a signal of a fixed spatiotemporal scale (1.0 c/deg gabor subtending 3 deg of visual angle, lasting 100 ms), we estimated both photon and cortical noises by making measurements at low (0.8 cd/m^2) and high (80 cd/m^2) display luminances. For thresholds measured on a noise background, Gaussian dynamic display noise started and terminated 500 ms before and after signal presentation, and was of a magnitude sufficient to elevate contrast threshold by at least a factor of 2 relative to blank-field thresholds. Contrast thresholds measured in high levels of display noise also allow us to compute visual efficiency relative to an ideal observer model. Pupil size was measured under task conditions with an infrared video-camera to estimate retinal illumination. In order to compare photon noise levels across observers, from the measured photon noise and retinal illumination we compute retinal transduction efficiency, which is illumination-independent. We compared our estimates of transduction efficiency, cortical noise and visual efficiency from snow patients, with those from 16 normal observers. Results: All of our patients reported fine-grained dynamic "snow" involving the entire visual fields of both eyes. Aside from refractive abnormalities in two patients, visual functions were otherwise normal based on a careful neuro-ophthalmological screen. All 5 patients reported the snow to be stronger at low light levels. However, visual efficiency, transduction efficiency, and cortical noise for all patients were statistically indistinguishable from the control group. Conclusions: Visual snow patients have normal levels of equivalent input noise, contrast sensitivity and visual efficiency. We propose that patients with visual snow have normal intrinsic visual noise but increased perceptual gain.

pdf
poster
Pelli, D. G. (2009). Towards an easier way to measure the visual span. [Abstract]. Journal of Vision, 9(8): 1002, http://journalofvision.org/ [Vision Sciences Society, Naples, Florida, May 9-14, 2009] pdf
poster Tillman, K. A., & Pelli, D. G. (2009). Reading pictures. [Abstract]. Journal of Vision, 9(8): 803, http://journalofvision.org/ [Vision Sciences Society, Naples, Florida, May 9-14, 2009] pdf
poster Rosen, S. (2009). A new technique for measuring the critical spacing of crowding. [Abstract]. Journal of Vision, 9(8): 998, http://journalofvision.org/ [Vision Sciences Society, Naples, Florida, May 9-14, 2009] pdf
  Chakravarthi, R., Tillman, K. A., & Pelli, D. G. (2009). Features used or features available? [Abstract]. Journal of Vision, 9(8): 789  
poster
Tillman, K. A., Araki, M., & Pelli, D. G. (2008). Crowding shows that faces have parts and bodies do not. Perception 37 ECVP Abstract Supplement, page 33. http://www.perceptionweb.com/abstract.cgi?id=v080115 pdf
poster Cantone, A. R., Tillman, K. A., & Pelli, D. G. (2008). Eccentric features integrate slowly [Abstract]. Journal of Vision, 8(6):653, http://journalofvision.org/8/6/653/ pdf
poster Suchow, J. W., & Pelli, D. G. (2008). Letter learning: Feature detection and integration [Abstract]. Journal of Vision, 8(6):1133, http://journalofvision.org/8/6/1133/ pdf
poster
Pelli, D., Song, S., & Levi, D. (2007). Amblyopic reading is crowded. [Abstract]. Journal of Vision, 7(9):519, http://journalofvision.org/7/9/519/, doi:10.1167/7.9.519. [Vision Sciences Society, Sarasota, Florida, May 2007] Published as Levi, Song, and Pelli (2007). pdf
poster S. Song, D. M. Levi, and D. G. Pelli (2007) Can “equivalent eccentricity” account for amblyopic vision? Association for Research in Vision and Ophthalmology. [Abstract] pdf
poster
Baron, J., & Pelli, D. G. (2006). Crowding counting. [Abstract]. Journal of Vision, 6(6):198. http://journalofvision.org/6/6/198 [Vision Sciences Society, Sarasota, Florida, May 2006]

 

pdf
poster Bergman, C., Martelli, M., Burani, C., Pelli, D. G., & Zoccolotti, P. (2006). How the word length effect develops with age. [Abstract]. Journal of Vision, 6(6):999. http://journalofvision.org/6/6/999 [Vision Sciences Society, Sarasota, Florida, May 2006] pdf
poster Pelli, D. G., & Tillman, K. A. (2006). Crowding limits reading. [Abstract]. Journal of Vision, 6(6):993, http://journalofvision.org/6/6/993/, doi:10.1167/6.6.993. [Vision Sciences Society, Sarasota, Florida, May 2006] Published as Pelli, Tillman, Freeman, Su, Berger, & Majaj (2007). pdf
poster Tillman, K. A., Pelli, D. G., Martelli, M., Stott, J., & Rosenblatt, J. (2006). Is reading serial? [Abstract]. Journal of Vision, 6(6):995, http://journalofvision.org/6/6/995/, doi:10.1167/6.6.995. [Vision Sciences Society, Sarasota, Florida, May 2006] pdf
poster
Oruç, I., Landy, M. S., & Pelli, D. G. (2005). Noise masking reveals channels for second-order letters. [Abstract]. Journal of Vision, 5(8): 183, http://journalofvision.org/5/8/183. [Vision Sciences Society, Sarasota, Florida, May 2005]
pdf Published as Oruç, I., Landy, M. S., & Pelli, D. G. (2006) Noise masking reveals channels for second-order letters. Vision Research, 46, 1493–1506. [PubMed]
pdf
poster Pelli, D. G., Su, M., Berger, T.D., Majaj, N.J., Martelli, M., Guo, S., & Tillman, K. (2005). Crowding, shuffling, and capitalizing reveal three processes in reading. [Abstract]. Journal of Vision, 5(8):806, http://journalofvision.org/5/8/806. [Vision Sciences Society, Sarasota, Florida, May 2005] Published as Pelli & Tillman (2007). pdf
poster Suchow, J. W., & Pelli, D. G. (2005). Learning to identify letters: Generalization in high-level perceptual learning. [Abstract]. Journal of Vision, 5(8):712, http://journalofvision.org/5/8/712. [Vision Sciences Society, Sarasota, Florida, May 2005] pdf
poster
Pelli, D. G., Martelli, M., and Majaj, N. J. (2004) Using crowding to determine whether an object is identified as a whole or by parts [Abstract]. Journal of Vision, 4(8):507, http://journalofvision.org/4/8/507. [Vision Sciences Society, Sarasota, Florida, May 2004] pdf
poster James, K. H., Martelli, M., James, T. W., Majaj, N. J., Pelli, D. G., and Gauthier, I. (2004) fMRI reveals the role of the left anterior fusiform gyrus in letter detection [Abstract]. Journal of Vision, 4(8):512, http://journalofvision.org/4/8/512. [Vision Sciences Society, Sarasota, Florida, May 2004] pdf
poster
Steingrimsson, R., Majaj, N. J., & Pelli, D. G. (2003) Where are letters processed and learned? Neural specialization for letter processing under different learning conditions. Perception 32 supplement. pdf
poster Steingrimsson, R., Gilman, J., Ramos, E., Majaj, N. J., & Pelli, D. G. (2003) Where are letters learned? An fMRI study. NYU Natural Sciences Poster Session. New York University, New York, NY, June 25, 2003.
pdf
poster Berger, T. D., Martelli, M., Su, M., Aguayo, M., Majaj, N. J., & Pelli, D. G. (2003). Reading quickly in the periphery [Abstract]. Journal of Vision, 3(9):806, http://journalofvision.org/3/9/806/, doi:10.1167/3.9.806. [Vision Sciences Society, Sarasota, Florida, May 2003] Published as Pelli, Tillman, Freeman, Su, Berger, & Majaj (2007). pdf
poster Majaj, N. J., Liang, Y. X., Martelli, M., Berger, T. D., & Pelli, D. G. (2003). Channel for reading [Abstract]. Journal of Vision, 3(9):813, http://journalofvision.org/3/9/813/, doi:10.1167/3.9.813. [Vision Sciences Society, Sarasota, Florida, May 2003] pdf
poster Martelli, M., Silla, S., Majaj, N. J., & Pelli, D. G. (2003). Complexity impairs efficiency in the periphery [Abstract]. Journal of Vision, 3(9):505, http://journalofvision.org/3/9/505/, doi:10.1167/3.9.505. [Vision Sciences Society, Sarasota, Florida, May 2003] pdf
poster Pelli, D. G., Martelli, M., Majaj, N. J., & Berger, T. D. (2003). One channel per object? [Abstract]. Journal of Vision, 3(9):267, http://journalofvision.org/3/9/267/, doi:10.1167/3.9.267. [Vision Sciences Society, Sarasota, Florida, May 2003] pdf
poster
Mishra, A., Baweja, G., Martelli, M., Chen, I., Fox, J., Majaj, N. J., & Pelli, D. G. (2002) How efficiency for identifying objects improves with age. European Conference on Visual Perception, Glasgow, August 2002. pdf
poster Pelli, D., Lee, M. H., Martelli, M., & Majaj, N. J. (2002).The orientation filter we use to identify objects: object recognition by a donut [Abstract]. Journal of Vision, 2(7):699, http://journalofvision.org/2/7/699/, doi:10.1167/2.7.699. [Vision Sciences Society, Sarasota, Florida, May 10-15, 2002] pdf
poster Martelli, M., Majaj, N. J., & Pelli, D. (2002). Words and faces: eccentricity distinguishes crowding from context [Abstract]. Journal of Vision, 2(7):608, http://journalofvision.org/2/7/608/, doi:10.1167/2.7.608. [Vision Sciences Society, Sarasota, Florida, May 10-15, 2002]

Published as Martelli, Majaj, and Pelli (2005).
pdf

Pelli, D. G., & Palomares, M. (2000). The role of feature detection in crowding. Investigative Ophthalmology and Visual Science, 41, S37.

ABSTRACT. A letter in the periphery is much harder to identify in the presence of neighboring letters. This is crowding. Last year we showed that crowding in utterly unlike ordinary masking, having a very steep contrast response and lacking channel-like selectivity (Palomares et al., ARVO '99). We now report that the effect of each flanking letter on the threshold contrast for identification of the target letter is all or none, dependent on target-flank separation but independent of suprathreshold flank contrast. This is evidence that the interference occurs at a second stage, between detected features.

 

Palomares, M., LaPutt, M. C., & Pelli, D. G. (1999). Crowding is unlike ordinary masking. Investigative Ophthalmology and Visual Science, 40, 1864.

ABSTRACT. Operationally, masking is the effect of a “mask” pattern on visibility of a signal. Usually the signal threshold is proportional (log-log slope of 1) to mask contrast, or nearly so (log-log slope of 0.65). And masks are usually effective only if they overlap the signal. In the periphery, small letters are much harder to identify in the presence of nearby letters. This is crowding. At 4° viewing eccentricity, we find that threshold contrast for identification of a 0.25° signal letter is elevated 10-fold by mask letters anywhere in a 2.5° region, ten times wider than the signal. Threshold is a sigmoidal function of flank contrast, with a log-log slope of 2. If the signal is instead masked by white noise, noise is effective only over the 0.25° region of the signal itself and threshold is proportional to mask contrast. These results and observers’ introspections suggest that the noise mask interferes with feature detection while crowding represents inappropriate combination of detected features.

 

Bayer, H. M., Schwartz, O., & Pelli, D. (1998). Recognizing facial expressions efficiently. Investigative Ophthalmology and Visual Science, 39, ?.

ABSTRACT. The roles of facial features and spatial frequencies in the recognition of facial expressions have been studied separately, but not together. We measured efficiency for observers performing a 21-way facial expression identification task. Baseline efficiency for the original photos was only 2%. We wondered if human performance would improve by limiting the differences between stimuli to a critical spatial frequency range and area as determined by Schwartz, et al. Limiting the differences to the critical spatial frequency band centered at 8 cycles/face width improved efficiency to 4%. When images differed only within a critical area that included the mouth and the edge of the cheek, efficiency increased to 6%. Finally, by limiting the differences to the critical spatial frequencies within the critical region, we reached an efficiency of 9%.

 
 

Palomares, M., Cardazone, G., Green, J., LaPutt, M. C., Majaj, N. J., Pelli, D. G., & Levi, D. M. (1998). Letter identification is channel mediated, but crowding isn't. Investigative Ophthalmology and Visual Science, 39, 833.

ABSTRACT. Crowding is often described as contour interaction, implying that the crowding effect is due to interference between neighboring features. Using critical-band masking, Majaj et al. (ARVO ‘97) showed that letter identification is mediated by a spatial frequency channel whose center frequency falls as the -0.7 power of letter size. We measured critical spacing: the minimum spacing between letters required to eliminate the effect of crowding. Interference within a channel ought to extend over a fixed number of periods of the channel’s center frequency, predicting that critical spacing should be proportional to the 0.7 power of letter size. Instead, we find that critical spacing is directly proportional to letter size, and identical in foveal, peripheral, and amblyopic vision. Replacing a letter S by its outline S doubles its line frequency, and doubles the channel frequency. Yet we find the same critical spacing for crowding of a letter by either regular or outline letters. The finding that crowding is characterized by the scale of the whole letter, rather than that of its features, indicates that crowding represents interference at the level of recognizing the whole pattern rather than detection of individual features.

 
 

Schwartz, O., Bayer, H. M., & Pelli, D. (1998). Features, frequencies, and facial expressions. Investigative Ophthalmology and Visual Science, 39, ?.

ABSTRACT. We find that only a particular range of spatial frequencies and spatial features is required to identify facial expressions (1 of 21). Critical spatial frequencies are obtained by varying the cut-off frequency of lowpass and highpass noise; critical features are obtained by varying the cut-off spatial postion of a horizontal and vertical noise curtain. The critical region in frequency and space is a band centered at 8 cycles per face width, stretching horizontally from the tip of the lips to the edge of the cheek and vertically from the bottom of the nose to the chin. The photo illustrates the critical information that we found. It is constructed by using one fixed face as the background and pasting onto it part of one of the 21 images, just in the required region and frequency range.

 
Pelli, D. G. (1997) Two stages of perception. Paper presented at Theories of Vision symposium at New York University, New York City, October 17, 1997. http://psych.nyu.edu/pelli/posters.html#1997 pdf
Raghavan, M., & Pelli, D. G. (1995). Psychophysical evidence for cortical noise. Investigative Ophthalmology and Visual Science, 36(4), s905.  
  Pelli, D. G., Raghavan, M., & Ahuja, S. (1995). The noises that limit visual perception. Investigative Ophthalmology and Visual Science, 36(4), s851.  
poster Cornelissen, F. W., Pelli, D. G., Farell, B., & Szeverenyi, N. (1995) fMRI of contrast response in visual cortex. Human Brain Mapping, 1 (Suppl. 1), 45. pdf
Pelli, D. G. (1981) The effect of uncertainty: detecting a signal at one of ten-thousand possible times and places. Supplement to Investigative Ophthalmology and Visual Science, 20, 178A. [Complete text and figures of the talk appear as Appendix 6 in my thesis: pdf]  
to the top

pdf pdf Need help downloading and opening a PDF file?

Updated