% % This file was created by the Typo3 extension % sevenpack version 0.7.14 % % --- Timezone: CEST % Creation date: 2013-05-22 % Creation time: 07-37-39 % --- Number of references % 21 % @Article { 4796, title = {Object features used by humans and monkeys to identify rotated shapes}, journal = {Journal of Vision}, year = {2008}, month = {2}, volume = {8}, number = {2:9}, pages = {1-15}, abstract = {Humans and rhesus monkeys can identify shapes that have been rotated in the picture plane. Recognition of rotated shapes can be as efficient as recognition of upright shapes. Here we investigate whether subjects showing view-invariant performance use the same object features to identify upright and rotated versions of a shape. We find marked differences between humans and monkeys. While humans tend to use the same features independent of shape orientation, monkeys use unique features for each orientation. Humans are able to generalize to a greater degree across orientation changes than rhesus monkey observers, who tend to relearn separate problems at each orientation rather than flexibly apply previously learned knowledge to novel problems.}, department = {Department Logothetis}, web_url = {http://journalofvision.org/8/2/9/Nielsen-2008-jov-8-2-9.pdf}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, language = {en}, DOI = {10.1167/8.2.9}, author = {Nielsen, KJ and Logothetis, NK and Rainer, G} } @Article { 4316, title = {Object Recognition: Similar Visual Strategies of Birds and Mammals}, journal = {Current Biology}, year = {2007}, month = {3}, volume = {17}, number = {5}, pages = {R174-R176}, abstract = {Behavioral testing has revealed that pigeons may use the same visual information sources as humans to discriminate between three-dimensional shapes.}, department = {Department Logothetis}, web_url = {http://www.sciencedirect.com/science?_ob=MImg\&_imagekey=B6VRT-4N689JT-K-3\&_cdi=6243\&_user=29041\&_orig=browse\&_coverDate=03\%2F06\%2F2007\&_sk=999829994\&view=c\&wchp=dGLbVtz-zSkWb\&md5=1af621df50c1268dd8c291426ace0718\&ie=/sdarticle.pdf}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, language = {en}, DOI = {10.1016/j.cub.2007.01.014}, author = {Nielsen, KJ and Rainer, G} } @Article { 4140, title = {Dissociation between LFP and spiking activity in macaque inferior temporal cortex reveals diagnosticity-based encoding of complex objects}, journal = {Journal of Neuroscience}, year = {2006}, month = {9}, volume = {26}, number = {38}, pages = {9639-9645}, abstract = {Neurons in the inferior temporal (IT) cortex respond selectively to complex objects, and maintain their selectivity despite partial occlusion. However, relatively little is known about how the occlusion of different shape parts influences responses in the IT cortex. Here, we determine experimentally which parts of complex objects monkeys are relying on in a discrimination task. We then study the effect of occlusion of parts with different behavioral relevance on neural responses in the IT cortex at the level of spiking activity and local field potentials (LFPs). For both spiking activity and LFPs, we found that the diagnostic object parts, which were important for behavioral judgments, were preferentially represented in the IT cortex. Our data show that the effects of diagnosticity grew systematically stronger along a posterior–anterior axis for LFPs, but were evenly distributed for single units, suggesting that diagnosticity is first encoded in the posterior IT cortex. Our findings highlight the power of co mbined analysis of field potentials and spiking activity for mapping structure to computational function in the brain.}, department = {Department Logothetis}, web_url = {http://www.jneurosci.org/cgi/reprint/26/38/9639}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, language = {en}, DOI = {10.1523/JNEUROSCI.2273-06.2006}, author = {Nielsen, K and Logothetis, NK and Rainer, G} } @Article { 3820, title = {Discrimination strategies of human and monkey for complex visual displays}, journal = {Current Biology}, year = {2006}, month = {4}, volume = {16}, number = {8}, pages = {814-820}, abstract = {By learning to discriminate among visual stimuli, human observers can become experts at specific visual tasks. The same is true for Rhesus monkeys, the major animal model of human visual perception. Here, we systematically compare how humans and monkeys solve a simple visual task. We trained humans and monkeys to discriminate between the members of small natural-image sets. We employed the “Bubbles” procedure [1] to determine the stimulus features used by the observers. On average, monkeys used image features drawn from a diagnostic region covering about 7\% ± 2\% of the images. Humans were able to use image features drawn from a much larger diagnostic region covering on average 51\% ± 4\% of the images. Similarly for the two species, however, about 2\% of the image needed to be visible within the diagnostic region on any individual trial for correct performance. We characterize the low-level image properties of the diagnostic regions and discuss individual differences among the monkeys. Our results reveal tha t monkeys base their behavior on confined image patches and essentially ignore a large fraction of the visual input, whereas humans are able to gather visual information with greater flexibility from large image regions.}, department = {Department Logothetis}, web_url = {http://www.sciencedirect.com/science?_ob=MImg\&_imagekey=B6VRT-4JS21WP-11-1\&_cdi=6243\&_user=29041\&_orig=browse\&_coverDate=04\%2F18\%2F2006\&_sk=999839991\&view=c\&wchp=dGLbVlb-zSkWz\&md5=08d41ae24c5163164abab077315779e5\&ie}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, language = {en}, DOI = {10.1016/j.cub.2006.03.027}, author = {Nielsen, KJ and Logothetis, NK and Rainer, G} } @Article { 3833, title = {Fixations in natural scenes: interaction of image structure and image content}, journal = {Vision Research}, year = {2006}, month = {3}, volume = {46}, number = {16}, pages = {2535-2545}, abstract = {Explorative eye movements specifically target some parts of a scene while ignoring others. Here we investigate how local image structure - defined by spatial frequency contrast - and informative image content - defined by higher order image statistics - are weighted for the selection of fixation points. We measured eye movements of macaque monkeys freely viewing a set of natural and manipulated images outside a particular task. To probe the effect of scene content, we locally introduced patches of pink noise into natural images, and to probe the interaction with image structure, we altered the contrast of the noise. We found that fixations specifically targeted the natural image parts and spared the uninformative noise patches. However, both increasing and decreasing the contrast of the noise attracted more fixations, and, in the extreme cases, compensated the effect of missing content. Introducing delusive patches from another natural image led to similar results. In all paradigms tested, the interaction bet ween scene structure and informative scene content was the same in any of the first six fixations on an image, demonstrating that the weighting of these factors is constant during viewing of an image. These results question theories, which suggest that initial fixations are driven by stimulus saliency whereas later fixations are determined by informative scene content.}, department = {Department Logothetis}, web_url = {http://www.sciencedirect.com/science?_ob=MImg\&_imagekey=B6T0W-4JHMFCW-1-B\&_cdi=4873\&_user=29041\&_orig=browse\&_coverDate=08\%2F31\%2F2006\&_sk=999539983\&view=c\&wchp=dGLbVtb-zSkWW\&md5=6cfbfebe283a94a24048473552a43f33\&ie=}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, language = {en}, DOI = {10.1016/j.visres.2006.02.003}, author = {Kayser, C and Nielsen, KJ and Logothetis, NK} } @Article { 3819, title = {Eye movements of monkey observers viewing vocalizing conspecifics}, journal = {Cognition}, year = {2006}, month = {1}, volume = {101}, number = {3}, pages = {515-529}, abstract = {Primates, including humans, communicate using facial expressions, vocalizations and often a combination of the two modalities. For humans, such bimodal integration is best exemplified by speech-reading – humans readily use facial cues to enhance speech comprehension, particularly in noisy environments. Studies of the eye movement patterns of human speech-readers have revealed, unexpectedly, that they predominantly fixate on the eye region of the face as opposed to the mouth. Here, we tested the evolutionary basis for such a behavioral strategy by examining the eye movements of rhesus monkeys observers as they viewed vocalizing conspecifics. Under a variety of listening conditions, we found that rhesus monkeys predominantly focused on the eye region versus the mouth and that fixations on the mouth were tightly correlated with the onset of mouth movements. These eye movement patterns of rhesus monkeys are strikingly similar to those reported for humans observing the visual components of speech. The data therefore suggest that the sensorimotor strategies underlying bimodal speech perception may have a homologous counterpart in a closely related primate ancestor.}, department = {Department Logothetis}, web_url = {http://www.sciencedirect.com/science?_ob=MImg\&_imagekey=B6T24-4J5C802-1-K\&_cdi=4908\&_user=29041\&_orig=search\&_coverDate=10\%2F31\%2F2006\&_sk=998989996\&view=c\&wchp=dGLbVzz-zSkzS\&md5=44ee60f125ff59f612771fae32e6f567\&ie=/sdarticle.pdf}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, language = {en}, DOI = {10.1016/j.cognition.2005.12.007}, author = {Ghazanfar, AA and Nielsen, KJ and Logothethis, NK} } @Inbook { 4317, title = {Using spikes and local field potentials to reveal computational networks in monkey cortex}, year = {2008}, month = {11}, pages = {350-362}, abstract = {Traditionally, neurophysiological investigations in awake non-human primates have largely focused on the study of single-unit activity (SUA), recorded extracellularly in behaving animals using microelectrodes. The general aim of these studies has been to uncover the neural basis of cognition and action by elucidating the relation between brain activity and behavior. This is true for studies in sensory systems such as the visual system, where investigators are interested in how SUA covaries with aspects of visually presented stimuli, as well as for studies in the motor system where SUA covariation with movement targets and dynamics are investigated. In addition to these SUA studies, there has been increasing interest in the local field potential (LFP), a signal that reflects aggregate activity across populations of neurons near the tip of the microelectrode. In this chapter, we will describe recent progress in our understanding of brain function in awake behaving monkeys using LFP recordings. We will show that the combination of recording the activity of single neurons and local populations simultaneously offers a particularly promising way to gain insight into cortical brain mechanisms underlying cognition and memory.}, department = {Department Logothetis}, web_url = {http://ebooks.cambridge.org/chapter.jsf?bid=CBO9780511541650\&cid=CBO9780511541650A024}, editor = {H{\"o}lscher, C. , M. Munk}, publisher = {Oxford University Press}, address = {Oxford, England}, booktitle = {Information processing by neuronal populations}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, language = {en}, ISBN = {978-0-521-87303-1}, DOI = {10.1017/CBO9780511541650.014}, author = {Nielsen, KJ and Rainer, G} } @Techreport { 2075, title = {Psychophysical comparison of synthesis algorithms for natural images}, year = {2003}, month = {12}, number = {119}, abstract = {In this study, we used three computational algorithms to compute basis sets for natural image patches, such that each patch could be synthesized as a linear combination of basis functions. The two biologically plausible algorithms non-negative matrix factorization (NMF) and sparsenet (SPN) were compared to standard principal component analysis (PCA). We assessed human psychophysical performance at identifying natural image patches synthesized using different basis set sizes in each of the algorithms. We also computed the reconstruction error, which represents a simple objective measure of synthesis performance. We found that the reconstruction error was a good predictor of human psychophysical performance. Performance was best for PCA, followed by NMF and SPN despite large differences in basis function characteristics. All algorithms were well able to generalize to represent novel natural image patches. When applied to white noise patches instead of natural images, PCA and SPN outperformed NMF. This shows that of the three algorithms the one that is least biologically plausible (PCA) actually supported best psychophysical performance, suggesting that in the present study it is low-level quality of reconstruction that is the main determinant of psychophysical performance.}, url = {http://www.kyb.tuebingen.mpg.de/fileadmin/user_upload/files/publications/pdf2075.pdf}, department = {Department Logothetis}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, institution = {Max Planck Institute for Biological Cybernetics}, author = {Nielsen, K and Logothetis, NK and Rainer, G} } @Poster { 5545, title = {Neural encoding of species dependent face-categories in the macaque temporal cortex}, journal = {Neuroforum}, year = {2007}, month = {4}, volume = {13}, number = {Supplement}, pages = {767}, abstract = {When perceiving a face, we can easily decide whether it belongs to a human or non-human primate. It is thought that face information is represented by neurons in the macaque temporal cortex. However, the precise encoding mechanisms used by these neurons remain unclear. Here we use face stimuli of humans, monkeys and monkey-human hybrids (morphs) to gain a better understanding of these mechanisms, in particular of the categorization of faces into different species, and how learning affects representation of these stimuli. We perform single cell and local field potential (LFP) recordings in the inferior-temporal (IT) cortex of the macaque brain during a fixation task. To investigate the perceptual effects of our stimuli and possible relations to the neural data, we conduct in parallel psychophysical experiments with human subjects. On preliminary results of 75 recorded cells in one animal, we found 66 visual responsive neurons. From them, 12 were tuned to faces ('face-cells') and 9 to other test objects (like a hand, clock, fruits, etc.). Six 'face-cells' prefer monkeys while just two prefer humans. Considering the population activity, monkey faces elicited in general higher firing rates on the population of neurons (independent of its category) than human faces. Additionally, these firing rates change gradually according to the human/monkey ratio of the morphed stimuli. After measuring the perceptual category boundary between monkeys and humans faces in our human subjects, we founded that it is shifted to the human side, independent of the method we use to measure it. Our preliminary cell recordings suggest that neural responses (firing rates) of some cells differentiate between monkey and human faces. Besides, the tuning curves of some neurons and the population correlate with the human-ratio of the morphed stimuli. Our psychophysical experiments confirm, on the one hand, the perceptual effect of our stimuli in which we manipulate the human-monkey ratio and, on the other hand showed a tendency of our subjects to set the category boundary between humans and monkeys closer to the human side. All these findings point to different mechanisms used by the brain to encode human and monkey faces, which seem to be clearly represented by neurons in the inferior-temporal cortex of the money brain.}, department = {Department Logothetis}, web_url = {http://www.neuro.uni-goettingen.de/nbc.php?sel=archiv}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {G{\"o}ttingen, Germany}, event_name = {31st G{\"o}ttingen Neurobiology Conference}, author = {Sigala A., GR and Nielsen, K and Logothetis, NK and Rainer, G} } @Poster { 5551, title = {Using 3-D human-monkey morphs to explore the boundaries of species dependent face-categories in humans}, journal = {Perception}, year = {2006}, month = {8}, volume = {35}, number = {ECVP Abstract Supplement}, pages = {207-208}, abstract = {Face perception has often been investigated with human faces differing in categories such as race or gender. Here, we investigate the perceptual border across species. We applied a method based on support vector machines to generate images of hybrid monkey - human faces (‘morphs‘) with different levels of human contribution. In the ‘explicit‘ experiment, we asked subjects to rate morphs at different morph levels as ‘humans‘ or ‘monkeys‘. We found that subjects rated the morphs as humans when they had a human contribution of at least 56\%±3\%. In the ‘implicit‘ experiment, we asked whether subjects could distinguish between successively presented morphs differing by ±10\% morph level from a morph centre. By varying the morph centre value from 10\% to 90\%, we were able to measure subject\&lsqu o;s sensitivity to detect species differences along the human - monkey continuum. We found that the sensitivity of subjects to detect species differences was highest when morphs had a human contribution of 65\%±3\%. In summary, the human - monkey boundary does not lie at the midpoint of the human - monkey continuum, but tends to be shifted towards the human side. Our results reveal an asymmetry in the perception of human - monkey morphed faces, which may be species-specific and/or due to expertise.}, department = {Department Logothetis}, web_url = {http://www.perceptionweb.com/abstract.cgi?id=v060412}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {St. Petersburg, Russia}, event_name = {29th European Conference on Visual Perception}, language = {en}, author = {Sigala, R and Koch, A and Nielsen, KJ and Logothetis, NK and Rainer, G} } @Poster { 5853, title = {Inferior temporal cortex during real world vision}, year = {2006}, month = {6}, pages = {74}, abstract = {Much of current visual neuroscience is performed using standardized procedures. Most notably, these generally include stimulus delivery using computer displays, the requirement of fixation, repeated performance of experimental conditions and lengthy conditioning of animals on tasks to allow for behavioral reports. Correlating neural responses with stimulus characteristics and behavior lies at the heart of systems neuroscience. These controlled conditions have many advantages, but at the same time can only represent an approximation of the processes that occur during real world vision. But how much are we missing under these constraints? Real world vision is characterized by eye movements in three dimensions as observers fixate and track objects in the environment. What are the characteristics of spike trains collected under such conditions and how do they differ from those collected during task performance. How much can be said about neural activity by applying the correlational approach to data acquired under these conditions? Does what we learn about neural activity and selectivity during task performance generalize to real world vision? To begin to address these questions, we have recorded extracellular activity of several inferior temporal cortex neurons simultaneously while monkeys viewed face and object stimuli presented on a computer monitor at the center of gaze during fixation. Then we record activity of the same neurons during interaction with a human experimenter, while measuring the monkeys’ eye position and recording the visual input using a camera. We compare about 5 minutes of activity collected during these two conditions. Preliminary results suggest many IT neurons were dynamically modulated during real world vision. Peak firing rates (eg at 200ms binwidth) tended to be greater during real world vision than during task performance. Some IT neurons showed markedly different interspike interval distributions in the two conditions. Our findings suggest that a dynamic three dimensional visual environment may be a useful tool for elucidating the function of visual neurons.}, department = {Department Logothetis}, web_url = {http://www.areadne.org/2006/}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {Santorini, Greece}, event_name = {AREADNE 2006: Research in Encoding and Decoding of Neural Ensembles}, language = {en}, author = {Sigala, R and Liebe, S and Nielsen, K and Logothetis, NK and Reiner, G} } @Poster { 3642, title = {The influence of behavioral relevance on object representation in inferior temporal cortex}, year = {2004}, month = {10}, volume = {34}, number = {751.4}, abstract = {Partial occlusion of objects is common in natural environments, and in most cases does not affect the identification of the occluded object. However, under certain conditions identification fails. The distinction between the two cases lies in which object parts remain visible through the occluder. We have recently shown for Rhesus monkeys performing a discrimination task on sets of natural images that the occlusion of specific image regions systematically influences behavioral performance. Here, we report on how these differences between image regions influence the responses of area IT neurons to partially occluded images. For four natural images, we determined three partially occluded variants that were reliably identified by the monkeys, and three variants that the monkeys failed to identify correctly. Images and their occluded variants were specific for each monkey observer. The three variants of each category differed in the image portion visible through the occluder (10, 30 or 50\%). We recorded from 341 IT neurons in two monkeys. 102 neurons were selected because their preferred non-occluded image evoked responses significantly different from baseline (p<.01). To assess the influences of occlusion category for these neurons, we compared the responses to all six occluded variants of the preferred image using a two-way ANOVA (factors occlusion category, stimulus size). 33 neurons reliably distinguished between the occlusion categories at least at one size of visible image portion (p<.01). In 27 of the 33 neurons, firing rates were elevated when identifiable image portions were shown to the monkeys. These findings indicate that image portions that monkeys preferentially relied on to guide their behavior had a greater impact on the response of IT neurons than image portions that were determined to be less relevant for behavior.}, department = {Department Logothetis}, web_url = {http://www.sfn.org/absarchive/}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {San Diego, CA, USA}, event_name = {34th Annual Meeting of the Society for Neuroscience (Neuroscience 2004)}, author = {Nielsen, KJ and Logothethis, NK and Rainer, G} } @Poster { NielsenLG2004, title = {Eye movements of monkey perceivers during viewing of species-specific vocal signals}, year = {2004}, month = {6}, volume = {5}, number = {64}, abstract = {Primates use auditory and visual cues to process the vocal communication signals produced by members of their species. For human speech, facial cues are known to enhance perception of speech under noisy conditions. That is, if speech has to be followed at high background noise levels, the ability to see a talker’s face enhances intelligibility. Recently, studies have examined the eye movement patterns of humans while viewing talkers under varying levels of naturalistic background noise (Vatikiotis-Bateson et al. (1998) Perception \& Psychophysics 60: 926). Under all noise conditions, human perceivers mostly fixate the eye region, followed by the mouth region, but as the background noise increases, they increase their number of fixations on the mouth region. To characterize potential behavioral homologies between monkey and human vocal communication, we have been investigating the eye movement patterns of rhesus monkeys while they view digitized videos of conspecifics producing vocalizations under three different background noise levels. We are analyzing how the simultaneous presentation of a dynamic facial expression and its corresponding vocalization influences the gaze patterns of monkey observers. Such data will give us insights into the evolution of sensory and motor mechanisms used in primate vocal communication.}, department = {Department Logothetis}, web_url = {http://imrf.mcmaster.ca/IMRF/2004/}, event_place = {Barcelona, Spain}, event_name = {5th International Multisensory Research Forum (IMRF 2004)}, author = {Nielsen, KJ and Logothetis, NK and Ghazanfar, AA} } @Poster { 2566, title = {Influence of Occlusion on the Responses of Area TE Neurons in the Macaque Monkey}, year = {2004}, month = {2}, volume = {7}, pages = {140}, abstract = {Identifying an image presented behind an occluder is in many cases easily possible. However, under certain occlusion conditions, identication fails. Critically, identication performance depends on which image parts are visible through the occluder. Area TE in the macaque visual cortex is thought to play an important role in object recognition processes. Here, we systematically test how the occlusion of different image regions affected the responses of area TE neurons. Two monkeys (Macaca mulatta) learned to identify members of sets of natural images. We then used Bubbles to assess how the occlusion of different image regions inuenced an observers' performance in the identication task. Most importantly, we determined which image regions had to be visible to allow the observer to identify the image correctly (the “informative” image regions). In most cases, the visibility of a limited portion of the image consistently inuenced the monkeys' performance. Based on these results, we constructed observer-specic image versions that contained informative or uninformative regions only. Recording from neurons in area TE, we compared the responses evoked by the full images to the ones evoked by the image's informative or uninformative parts. Preliminary results suggest that informative regions had a greater inuence on the response and selectivity of TE neurons than non-informative regions. Thus, monkey observers tend to rely on restricted regions of complex natural scenes during identication tasks, and this has an impact on their representation in the inferior temporal cortex}, department = {Department Logothetis}, web_url = {http://www.twk.tuebingen.mpg.de/twk04/index.php}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {T{\"u}bingen, Germany}, event_name = {7th T{\"u}bingen Perception Conference (TWK 2004)}, author = {Nielsen, K and Logothetis, NK and Rainer, G} } @Poster { 2564, title = {Visual discrimination strategies of monkey and human observers}, year = {2003}, month = {11}, volume = {33}, number = {590.20}, abstract = {During visually guided behavior, relevant information must be extracted from the environment. Which particular information about a stimulus is utilized to perform a task can provide insights into the strategies underlying behavior. Here, we identified the discrimination strategies of monkey (Macaca mulatta) and human observers for different sets of stimuli. One set contained geometrical shapes, the other stimulus set consisted of photographs of natural scenes. Each member of a set was associated with a unique behavioral response. The observers were first taught to execute the correct response upon presentation of each of the stimuli. Relevant stimulus regions were then determined by presenting stimuli behind occluding surfaces with randomly placed windows. The observers continued to perform the discrimination task on these occluded stimuli. In this paradigm, the diagnostic value of a shape region is reflected in the way in which its occlusion affects the performance of the observer. While occlusion of an unimportant stimulus feature will leave discrimination performance unaffected, occlusion of the relevant features will render the observer unable to identify the stimulus correctly. Overall, discrimination strategies were very similar between monkey and human observers. However, the monkeys tended to use smaller regions of the images than the human observers. In addition, we analyzed the eye movement patterns evoked when viewing the same images. The results for the monkey observers demonstrated that relevant image regions were more likely to be fixated than other image regions. Similarly, in human observers, the number of fixations made on a given region correlated with the information content subjectively assigned to that region. In summary, we determined directly how human and monkey observers use stimulus information to perform a visual task. We are using this technique to study the neural representation of objects in macaque temporal cortex.}, department = {Department Logothetis}, web_url = {http://www.sfn.org/index.aspx?pagename=annualmeeting_futureandpast}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {New Orleans, LA, USA}, event_name = {33rd Annual Meeting of the Society for Neuroscience (Neuroscience 2003)}, author = {Nielsen, K and Logothetis, NK and Rainer, G} } @Poster { 2565, title = {Visual discrimination strategies of monkey and human observers for simple geometric shapes}, year = {2003}, month = {2}, volume = {6}, pages = {159}, abstract = {During visually guided behavior relevant information must be extracted from the objects in the environment. Which information about particular stimuli is actually used can provide insights into strategies underlying behavior. Here we quantify which parts of simple geometric shapes humans and monkey observers used during a visual discrimination task. The stimulus set consisted of three rotated versions of a square with a single protrusion (left, up or right). Each shape was associated with a unique behavioral response (button press or eye movement). We determined relevant regions in each stimulus by presenting the shapes behind occluding surfaces with randomly placed windows, while the observers continued to perform the discrimination task. In this paradigm, the diagnostic value of a shape region is reflected in the way in which its occlusion affects the performance of the observer. While occlusion of an unimportant shape feature will leave discrimination performance unaffected, occlusion of the diagnostic features will render the observer unable to identify the shape correctly. For the human observers, each pattern had one diagnostic region consisting of the protrusion. Similar results were obtained in a monkey subject for two of the three shapes. Interestingly, no region was identifiable for the third shape, even though the monkey could identify the shape with about the same accuracy as the other two shapes. We suggest that the lack of a diagnostic region for the third shape reflects the strategy of the monkey, who tended to choose the third shape whenever neither of the protrusions corresponding to the first two shapes was visible. We conclude that although there was close agreement for the majority of stimuli, we also found strategy differences between human and monkey observers.}, department = {Department Logothetis}, web_url = {http://www.twk.tuebingen.mpg.de/twk03/}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {T{\"u}bingen, Germany}, event_name = {6. T{\"u}binger Wahrnehmungskonferenz (TWK 2003)}, author = {Nielsen, K and Logothetis, NK and Rainer, G} } @Poster { 1594, title = {Studying the representation of natural images using behavioural reverse correlation}, journal = {Perception}, year = {2002}, month = {8}, volume = {31}, number = {ECVP Abstract Supplement}, pages = {134}, abstract = {What information do humans use during the identification of complex natural scenes? To address this question, we employed a technique described originally by Gosselin and Schyns (2001 Vision Research 41 2261 - 2271), which identifies image regions diagnostic for visual-recognition tasks. We applied this method to a task where subjects had to discriminate natural images. On each trial, one of four images was shown behind an occluding mask punctured by multiple randomly located Gaussian windows. Diagnostic image regions were computed by comparing masks that resulted in correct performance with masks leading to incorrect performance. During different sessions, we used either a constant-stimuli protocol with a fixed number of windows, or a staircase protocol to adjust the number of windows as a function of behavioural performance. In general, depending on the particular natural image, different regions were revealed as diagnostic. Results for the constant-stimuli and the staircase protocols were in good agreement. For the constant-stimuli protocol, we found that subjects' behavioural performance improved with training for some natural images. Diagnostic information generally did not show dramatic changes, although sometimes particular image regions became diagnostic with learning. These results demonstrate that reverse correlation can be used to reveal diagnostic regions in complex natural images.}, department = {Department Logothetis}, web_url = {http://www.perceptionweb.com/abstract.cgi?id=v020429}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {Glasgow, UK}, event_name = {25th European Conference on Visual Perception}, author = {Nielsen, K and Rainer, G and Brucklacher, V and Logothetis, NK} } @Poster { 968, title = {Comparison of image decomposition techniques: results from psychophysics and computation}, journal = {Perception}, year = {2001}, month = {8}, volume = {30}, number = {ECVP Abstract Supplement}, pages = {108}, abstract = {Our visual environment is highly structured, generating complex statistical dependencies in natural scenes. In the framework of information theory, these dependencies of input variables generate redundancy in the input. Removing the redundancies would lead to the construction of an efficient code for natural scenes. A number of computational algorithms have been developed that are based on the assumption of redundancy reduction. Four of these algorithms were used here that are often discussed in the context of natural-scene statistics. These algorithms were principal component analysis, independent component analysis, sparsenet, and non-negative-matrix factorisation. Each of the algorithms decomposes the natural scenes into basis features or functions. The input can then be reconstructed again by a linear combination of these basis functions. Usually, the performance of these decomposition algorithms is evaluated by calculating analytically defined measures, eg the reconstruction error between the original and the reconstructed image or the redundancy reduction achieved with the decomposition. Here, in addition to the calculation of reconstruction errors, we tested the algorithms in three psychophysical experiments. In each of the experiments, subjects had to match reconstructed images to their originals in a delayed match-to-sample paradigm. The performance of subjects in this task depends on the ability of the algorithms to preserve the information present in the original scenes. The three experiments tested different properties of the algorithms. The first experiment was concerned with the dependence of the psychophysical performance on the number of basis functions used in the reconstruction. Since each of the algorithms derives its basis functions by applying a statistical criterion to a set of training images, it is an important question how general these basis functions are. This was tested in the second experiment. Finally, the third experiment assessed the robustness of the algorithms against noise added in the reconstruction process. This evaluation of decomposition algorithms in a more natural context provides new insights that can extend the computational results. It also helps to clarify which statistical criteria are of importance in natural vision as opposed to purely computational purposes.}, department = {Department Logothetis}, web_url = {http://www.perceptionweb.com/abstract.cgi?id=v010224}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {Kusadasi, Turkey}, event_name = {Twenty-fourth European Conference on Visual Perception}, author = {Nielsen, K and Rainer, G and Logothetis, NK} } @Poster { NielsenRL2001, title = {Comparison of image decomposition techniques: results from psychophysics and computation}, year = {2001}, month = {3}, pages = {176}, abstract = {Our visual environment is highly structured, generating complex statistical dependencies in natural scenes. In the framework of information theory, these dependencies of input variables generate redundancy in the input. Removing the redundancies would lead to the construction of an efficient code for natural scenes. A number of computational algorithms have been developed that are based on the assumption of redundancy reduction. Four of these algorithms were used here that are often discussed in the context of natural scene statistics. These algorithms were Principal Component Analysis, Independent Component Analysis, Sparsenet and Non-Negative-Matrix Factorization. Each of the algorithms decomposes the natural scenes into basis features or functions. The input can then be reconstructed again by a linear combination of these basis functions. Thus, the decomposition algorithms can be seen as performing filtering of the input data, such that the basis functions filter statistical dependencies. Because each of the four algorithms has a different statistical criterion for constructing the basis functions, each of them filters different types of statistics in the natural images. Usually, the performance of these decomposition algorithms is evaluated by calculating analytically defined measures, e.g. the reconstruction error between the original and the reconstructed image or the redundancy reduction achieved with the decomposition. Here, in addition to the calculation of reconstruction errors, we tested the algorithms in three psychophysical experiments. In each of the experiments, subjects had to match reconstructed images to their originals in a delayed match-to-sample paradigm. The performance of subjects in this task depends on the ability of the algorithms to preserve the information present in the original scenes. The three experiments tested different properties of the algorithms. The first experiment was concerned with the dependence of the psychophysical performance on the number of basis functions used in the reconstruction. Since each of the algorithm derives its basis functions by applying a statistical criterion to a set of training images, it is an important question how general these basis functions are. This was tested in the second experiment. Finally, the third experiment assessed the robustness of the algorithms against noise added in the reconstruction process. This evaluation of decomposition algorithms in a more natural context provides new insights that can extend the computational results. It also helps to clarify which statistical criteria are of importance in natural vision as opposed to purely computational purposes.}, department = {Department Logothetis}, web_url = {http://www.twk.tuebingen.mpg.de/twk01/Pcomp.htm}, event_place = {T{\"u}bingen, Germany}, event_name = {4. T{\"u}binger Wahrnehmungskonferenz (TWK 2001)}, author = {Nielsen, K and Rainer, G and Logothetis, NK} } @Conference { 5000, title = {Decoding the human/monkey face category boundary from the macaque inferior-temporal (IT) cortex using 3D human/monkey morphed faces}, year = {2007}, month = {11}, volume = {37}, number = {554.8}, abstract = {Ambiguous stimuli constitute a powerful method to dissociate between the physical properties of the stimuli and their representation in the brain. Following this idea, we applied a new computer-vision algorithm based on Support-Vector-Machines (SVMs) to create three-dimensional morphed faces (linear interpolated) between humans and monkeys in order to investigate how species-dependent face information is encoded in the inferior-temporal (IT) cortex of the macaque brain. Previous psychophysical experiments using these stimuli have shown that human subjects tend to classify ambiguous morphs as discrete instances of the human/monkey categories (‘categorical perception’). Moreover, subjects draw the category boundary closer to their own species (at approximately 60\%human/40\% monkey). We recorded the single-unit-activity (SUA) of 118 neurons and the local field potential (LFP) at 58 sites of the IT cortex of one macaque monkey during fixation of these morphed stimuli. Out of a total of 118 single units, 85\% were visually responsive, 23\% were selective to faces, 12\% selective to monkeys and 14\% to humans, according to standard criteria. To analyze the population activity, we trained different classifiers (k-Nearest Neighbor, Support vector Machines, K-Means) to learn the representation (SUA and LFPs) of human and monkey faces and tested them with the ambiguous stimuli. We found that, symmetric to the findings in humans, ambiguous faces are categorized by the pattern classifiers in a manner implying a categorical representation of the faces. Furthermore, the classifiers drew the category boundary closer to the monkey category (at approximately 40\%human/60\% monkey) for both kinds of neural signals. In contrast to the linear change of the morphed faces, our preliminary results showed that the neural representation of the species information is nonlinear. This nonlinearity suggests an ‘own-species’ advantage in the encoding of face stimuli. Consistent with learning theories, this advantage seems to be better reflected in our data by a sharper tuning of the monkey-selective cells compared to the human-selective, and not by a difference in the number of cells.}, department = {Department Logothetis}, talk_type = {Abstract Talk}, web_url = {http://www.sfn.org/am2007/}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {San Diego, CA, USA}, event_name = {37th Annual Meeting of the Society for Neuroscience (Neuroscience 2007)}, language = {en}, author = {Sigala Alanis, GR and Nielsen, KJ and Logothetis, NK and Rainer, G} } @Conference { 3711, title = {Local field potentials recorded from macaque TE reveal posterior to anterior gradient in the extraction of diagnostic elements from natural scenes}, year = {2005}, month = {11}, volume = {35}, number = {362.2}, abstract = {Partial occlusion of objects is ubiquitous in nature. We have shown that Rhesus monkeys are able to identify partially occluded stimuli as long as diagnostic image parts remain visible. Based on these results, we constructed partially occluded variants for a number of natural scenes, so that the scenes’ diagnostic parts were either visible or occluded. Here we report how the local field potentials (LFPs) recorded in area TE depend on the diagnosticity of a visible image region. The LFP was recorded from a total of 214 sites in two Rhesus monkeys. We sampled a large extent of area TE ranging from AP 11 to 23 mm, and covering the lower bank of the STS and lateral TE. LFPs were characterized by computing the visual evoked potentials (VEPs). In both monkeys, we found sites at which diagnosticity was a significant determinant of VEP amplitude. Interestingly, these sites were distributed according to a posterior to anterior gradient, and were more common in anterior TE (\(\chi\)2-test, p<.001). We also examined how diagnosticity modulated the raw LFP amplitudes. Again, the influence of diagnosticity on the LFP increased from posterior to anterior recording locations (correlation with recording location, p<.01). An analysis of single neuron responses recorded at the same sites revealed no significant differences along this posterior to anterior axis (correlation with recording location, p=.59). Our results reveal a posterior to anterior gradient in LFP amplitudes reflecting the extraction of diagnostic image parts. Such effects were not seen in single neuron responses collected at these sites. Because LFPs provide an estimate of local dendritic processing, our findings are consistent with the idea that diagnosticity is first encoded in posterior TE, but not in the preceding visual areas. These findings suggest that LFPs provide information about the distribution of function in the ventral pathway that is not available from single neuron activity.}, department = {Department Logothetis}, talk_type = {Abstract Talk}, web_url = {http://www.sfn.org/absarchive/}, institute = {Biologische Kybernetik}, organization = {Max-Planck-Gesellschaft}, event_place = {Washington, DC, USA}, event_name = {35th Annual Meeting of the Society for Neuroscience (Neuroscience 2005)}, digital = {1}, author = {Nielsen, KJ and Logothetis, NK and Rainer, G} }