Thesis (Ph.D.)--University of Florida, 2004. / Title from title page of source document. Document formatted into pages; contains 96 pages. Includes Vita. Includes bibliographical references.

The computational processes in the intermediate stages of the ventral pathway responsible for visual object recognition are not well understood. A recent physiological study by A. Pasupathy and C. Connor in intermediate area V4 using contour stimuli, proposes that a population of V4 neurons display bjectcentered,position-specific curvature tuning [18]. The “standard model” of object recognition, a recently developed model [23] to account for recognition properties of IT cells (extending classical suggestions by Hubel, Wiesel and others [9, 10, 19]), is used here to model the response of the V4 cells described in [18]. Our results show that a feedforward, network level mechanism can exhibit selectivity and invariance properties that correspond to the responses of the V4 cells described in [18]. These results suggest howobject-centered, position-specific curvature tuning of V4 cells may arise from combinations of complex V1 cell responses. Furthermore, the model makes predictions about the responses of the same V4 cells studied by Pasupathy and Connor to novel gray level patterns, such as gratings and natural images. Thesepredictions suggest specific experiments to further explore shape representation in V4.

In this paper, we introduce a novel set of features for robust object recognition, which exhibits outstanding performances on a variety ofobject categories while being capable of learning from only a fewtraining examples. Each element of this set is a complex featureobtained by combining position- and scale-tolerant edge-detectors overneighboring positions and multiple orientations.Our system - motivated by a quantitative model of visual cortex -outperforms state-of-the-art systems on a variety of object imagedatasets from different groups. We also show that our system is ableto learn from very few examples with no prior category knowledge. Thesuccess of the approach is also a suggestive plausibility proof for aclass of feed-forward models of object recognition in cortex. Finally,we conjecture the existence of a universal overcompletedictionary of features that could handle the recognition of all objectcategories.

O estudo da acuidade Vernier (VRN) revela a capacidade do sistema visual humano em detectar deslocamentos espaciais de poucos arcos de segundos, menores que a distância entre dois cones foveais adjacentes. Tal fato desperta interesse teórico sobre o tema, além de futuras aplicações na área clínica. A acuidade VRN pode ser medida tanto psicofisicamente quanto eletrofisiologicamente. Para a detecção de quebras de colinearidade (acuidade VRN), alguns autores hipotetizam que as células ganglionares (CGs) M da retina provêem sinal adequado da retina ao córtex, e dão suporte ao desempenho psicofísico da tarefa VRN. Em condições de estímulos semelhantes, as células ganglionares magnocelulares (M) em primatas parecem ter precisão espacial com razão sinal-ruído mais alta do que as células parvocelulares (P) . A dependência ao contraste (C) das células M na precisão espacial, frequência espacial, frequência temporal e velocidade do estímulo é mais similar ao desempenho psicofísico em humanos do que comparados aos dados das células P (Rüttiger et al., 2002; Sun et al., 2004). Nós utilizamos o Potencial Provocado Cortical Visual de Varredura (sVEP) para avaliar esta hipótese no nível de processamento intermediário entre as respostas de célula única na retina e a detecção psicofísica. Nós medimos os limiares corticais VRN em função do contraste (14 participantes, média de 28,21 ± 2,8) e lacunas (9 participantes, média de 29,7 ± 5,9). As quebras verticais VRN na colinearidade foram introduzidas em uma grade de onda quadrada horizontal. O estímulo VRN alternou entre um estado alinhado (grades sem quebras) e desalinhado (grades com quebras) a 6 Hz. Durante cada uma das 10 tentativas, o deslocamento aumentou em passos logarítmicos iguais de 0,5 a 7,5. O limiar VRN foi definido no momento do deslocamento em que a extrapolação linear da média vetorial das respostas em 1F atinge zero uV. Os contrastes testados foram: 4, 8, 16, 32, 64, 80%. Os resultados mostram que (1) aos limiares VRN em Log, medidos com sVEP, com o C em Log, diminuíram de forma linear (com uma inclinação de -0,5), similiares às células ganglionares M mas não P (Sun et al., 2004) e próximo às medidas psicofísicas (Sun et al., 2004; Wehrhahn e Westheimer, 1990); (2) Para C 16% obtivemos limiares de hiperacuidade (menor que 1 arcmin). Em altos contrastes a média do limiar foi de 0,37(erro padrão de 0,06 unidades logarítmicas); (3) Os limiares para o 2F tiveram uma dependência para o contraste diferente, com poucos efeitos para contraste abaixo de 16%. (4) As inclinações das linhas de extrapolação dos sVEP para o 1F1 foram 2 a 3 vezes maiores que as inclinações para 2F; (5) No protocolo controle, deslocamentos bidirecionais e simétricos geraram somente respostas no 2F. Os resultados 3 a 5 implicam que os componentes 1F e 2F derivam de neurônios distintos e fundamentam que respostas no 2F refletem respostas de movimento cortical simétrico. A dependência dos limiares de contraste do sVEP VRN (1F) é similiar aos estudos prévios psicofísicos (Sun et al., 2004; Wehrhahn e Westheimer, 1990), e repete a dependência ao contraste das células M (Sun et al., 2004). Estes resultados fundamentam a hipótese que o córtex extrai informações da posição relativa com precisão de hiperacuidade dos sinais advindos das células M / The human visual system is able to detect spatial displacements of a few arcsec, much smaller than the distance between two adjacent foveal cones. Hyperacuity tasks such as Vernier (VRN) have both theoretical importance as well as clinical application. VRN can be measured psychophysically and with sVEP. Some authors hypothesize that M ganglion cells provide the retinal signal to cortex adequate to support Vernier performance. Under stimulus conditions analogous to detection of Vernier offsets, primate magnocellular (M) ganglion cells appear to have more precise spatial localization (with higher Signal to Noise Ratio) than parvocellular (P) cells, and the dependence of M cell spatial precision on contrast (C), spatial frequency, temporal frequency and stimulus velocity is more similar to human psychophysical performance than comparable data from P cells (Ruttiger et al, 2002; Sun et al., 2003, 2004) (Rüttiger et al., 2002; Sun et al., 2004). We measured the C-dependence of cortical VRN thresholds (thd) using the Sweep VEP (sVEP) to help evaluate this hypothesis at a processing level intermediate between single-cell retinal responses and psychophysical detection. We measured Vernier thds using sVEP as function of constrast (12 young adults, age means 28.21 yrs ± 2.8) and Gaps (9 participants, 29.7 ± 5.9) with normal vision. Vertical VRN breaks in colinearity were introduced to a horizontal squarewave grating. The VRN stimulus alternated between aligned (grating w/o breaks) and misaligned (w/breaks) states at 6 (or 10) Hz. During each of ten, 10-s trials, displacement (D) was increased in equal logarithmic steps from 0.5 to 7.5. Vernier thd was defined as the D at which the rising slope of the vector averaged 1F response extrapolated to zero V. The Cs tested were: 4, 8, 16, 32, 64, 80%. We Found: (1) Log Vernier thd measuered with sVEP decreased linearly with log C similar to M- (but not P-) ganglion cells (Sun et al., 2004) with a slope of -0.5, close to that measured psychophysically (Rüttiger et al., 2002; Sun et al., 2004); (2) For C 16% , thds were true hyperacuities (less than 1). At high C, mean thd was 0.37(S.E = 0.06 log units); (3) Thds for 2F had a different C dependence, with little effect of C below 16 %. Thds for 2F were < 1F thds below 16 % C, but were 1F thds beyond 16 %; (4) The slopes of the sVEP extrapolation lines for 1F were 2-3 times > 2F slopes; (5) In a control protocol, symmetric, bidirectional displacements only generated 2F responses. Results 3-5 imply that the 1F and 2F components derive from distinct neurons, and support the notion that 2F responses reflect symmetric cortical motion responses. The C-dependence of sVEP Vernier (1F) thresholds is similar to prior psychophysics (Sun et al., 2004; Wehrhahn e Westheimer, 1990), and recapitulates Mcell C-dependence (Sun et al., 2004). This results support the hypothesis that cortex extracts relative position information with hyperacuity precision preferentially from M cell signals

Precise point-to-point connectivity is the basis of ordered maps of the visual field in the brain. One point in the visual field is represented at one locus in the dLGN and one locus in primary visual cortex. A fundamental problem in the development of most sensory systems is the creation of the topographic projections which underlie these maps. Mechanisms ranging from ordered ingrowth of fibres, through chemical guidance of axons to sculpting of the map from an early exuberant input have been proposed. However, we know little about how ordered maps are created beyond the first relay. What we do know is that a topological mismatch requires the exchange of neighbours in the geniculo-cortical projection and that manipulating the input to the primary relay can affect the geniculo-cortical topography. Taking advantage of the immaturity of the newborn hamster’s visual system, I studied the generation of an ordered map in primary visual cortex during the time of target innervation in normal and manipulated animals. I also investigated the patterning of neuronal activity prior to natural eye-opening. Paired injections of retrograde fluorescent tracers into visual cortex reveal that geniculate fibres are highly disordered at the time of invasion of the cortical plate. Topography in the geniculo-cortical projection emerges out of an unordered projection to area 17 in the first postnatal week. Furthermore, I show that manipulating the peripheral input can alter the topographic map which arises out of the early scatter. Removal of one eye at birth appears to slow the process of geniculo-cortical map formation ipsilateral to the remaining eye and at the end of the second postnatal week, a double projection between thalamus and cortex has formed. If retinal activity is blocked during this time, this double projection does not emerge. The results implicate retinal activity as the signal that induces the development of a different topographic order in the geniculo-cortical projection. It is generally believed that visual experience can influence development only after eye-opening. However, the final part of my thesis shows that neurons in the developing visual cortex of the ferret can not only be visually driven at least 10 days before natural eye-opening, but are also selective for differently oriented gratings presented <i>through the closed eye-lid</i>. Thus, visually-driven neuronal activity could influence development much earlier than previously assumed in many developmental studies.

Despite its great complexity, a great deal is known about the organization and information-processing properties of the visual system. However, the neural correlates of visual awareness are not yet understood. By studying patients with blindsight, the primary visual cortex (V1) has attracted a lot of attention recently. Although this brain area appears to be important for visual awareness, its exact role is still a matter of debate. Interactive models propose a direct role for V1 in generating visual awareness through recurrent processing. Hierarchal models instead propose that awareness is generated in later visual areas and that the role of V1 is limited to transmitting the necessary information to these areas. Interactive and hierarchical models make different predictions and the aim of this thesis is to review the evidence from lesions, perceptual suppression, and transcranial magnetic stimulation (TMS), along with data from internally generated visual awareness in dreams, hallucinations and imagery, this in order to see whether current evidence favor one type of model over the other. A review of the evidence suggests that feedback projections to V1 appear to be important in most cases for visual awareness to arise but it can arise even when V1 is absent.

Visual target acquisition is performed during several daily tasks, often requmng time¬dependent behavioural responses towards stimuli. Information processing during such tasks is subject to bottom-up as well top-down influences, which results in an integrated processing mechanism. It follows that if the underlying neural mechanisms can be elucidated, behaviour towards visual stimuli will be better understood, allowing for the development of visual environments that facilitates desired behavioural response. The current study aimed to develop a systems-level approach according to which the mechanisms that underlie visual target acquisition can be understood, by interpreting psychophysical data in terms of the structural and functional organization of the visual system. Empirical work entailed psychophysical experiments and elaborated on previous studies regarding conspicuity areas around and response time towards visual targets. The rationale was that these two measures can be used as an indication of the conspicuity of a target within a specific background, which in turn can be related to the nature of information processing during a target acquisition task. Results showed that a proportional relationship exists between the size of the conspicuity area and a target's perceived conspicuity, with the most conspicuous targets being associated with the largest conspicuity areas. Response time trends showed that target detectability at different positions within the conspicuity area is equal, but that detection performance at positions outside the conspicuity area is greatly influenced by the nature of the background surrounding the target. Interpretation of the results points to the importance of visual attention during target acquisition, which in turn is supported by the structural and functional organization of the visual system. Findings from the psychophysical study presented here, along with the proposed framework of information processing, emphasise that behavioural outcome during visual target acquisition cannot be explained without considering the structural and functional organization of the visual system. / Dissertation (MSc (Human Physiology))--University of Pretoria, 2006. / Physiology / unrestricted

Performance in visual tasks improves with attention, and this improvement has been shown to stem, in part, from changes in sensory processing. However, the mechanism by which attention affects perception remains unclear. Considering that neurons within the visual areas are selective for basic image statistics, such as orientation or spatial frequency (SF), it is plausible that attention modulates these sensory preferences by altering their so-called ‘tuning curves’. The goal of this project is to investigate this possibility by measuring and comparing the SF tuning curves across a range of attentional states in humans. In Experiment 1, a model-driven approach to fMRI analysis was introduced that allows for fast and efficient estimation of population spatial frequency tuning (pSFT) for individual voxels within human visual cortices. Using this method, I estimated pSFTs within early visual cortices of 8 healthy, young adults. Consistent with previous studies, the estimated SF optima showed a decline with retinotopic eccentricity. Moreover, my results suggested that the bandwidth of pSFT depends on eccentricity, and that populations with lower SF peaks possess broader bandwidths. In Experiment 2, I proposed a new visual task, coined the Numerosity Judgement Paradigm (NJP), for fine-grained parametric manipulation of attentional load. Eight healthy, young adults performed this task in an MRI scanner, and the analysis of the BOLD signal indicated that the activity within the putative dorsal attention network was precisely modulated as a function of the attentional load of the task. In Experiment 3, I used the NJP to modulate attentional load, and exploited the model-based approach to estimate pSFTs under different attentional states. fMRI results of 9 healthy, young adults did not reveal any changes in either peak or the bandwidth of the pSFTs with attentional load. This study yields a full visuocortical map of spatial frequency sensitivity and introduces a new paradigm for modulating attentional load. Although under this paradigm I did not find any changes in SF preferences within human visual areas with attentional load, I cannot preclude the possibility that changes emerge under different attentional manipulations.

While three decades of intensive cortical electrophysiology using a variety of sustained visual stimuli has made a significant contribution to many aspects of visual function, it has not supported the existence of intracortical circuit operations in cortical processing. This study investigated cortical processing by a comparison of the response of primary visual cortical neurones to transient electrical and strobe-flash stimulation. Experiments were performed on 74 anaesthetised Long Evans rats. Standard stereotaxic and extracellular electrophysiological techniques were employed. Continuous (on-line) raster plots and peri-stimulus time histograms (PSTHs) of the extracellular spikes from 81 visual cortical and 55 lateral geniculate nucleus (LGN) neurones were compiled. The strobe-flash stimuli (0.05 ms) were applied to the contralateral eye while the monopolar or bipolar electrical stimuli (0.2 ms, 80-400 μA) were applied to the ipsilateral LGN. 60 of the 81 (74%) tested cortical units were found to be responsive to visual stimuli. A distinct and consistent difference in the cortical response to the two types of transient stimuli was found: (a) Electrical stimulation evoked a prolonged period (197 ± 61 ms) of inhibition in all cortical neurones tested (n=20). This was the case even in those cortical units that were completely unresponsive to visual stimulation. The protracted inhibition was usually followed by a 100-200 ms phase of rebound excitation. (b) Flash stimulation evoked a prominent excitatory discharge (5-30 ms duration) after a latency of 30-60 ms from the onset of the stimulus (n = 59). This was followed by either moderate inhibition or return to a firing rate similar to control activity, for a maximum of 40 ms. Thereafter, cortical neurones showed a sustained increased level of activity with superimposed secondary excitatory phases. The duration of this late re-excitatory phase was 200-300 ms. In 17 of 20 (85%) tested units, the temporal profile of the cortical response to flash stimulation was modulated by small changes in the level of background illumination. In 16 of the 17 units, this sensitivity was reflected primarily as an emergence of a brief secondary inhibitory phase at the lowest level of background illumination (0 lux). Only 1 of the 17 cortical units displayed a flash-evoked primary inhibitory phase at O lux. We explored the possibility that neurones in the lateral geniculate nucleus (LGN) of the thalamus were responsible for the late phase of cortical reexcitation. 49 of the 55 (89%) LGN neurones could be classified as either of the "ON type" i.e. excited by visual stimuli, or the "OFF type" i.e. inhibited by visual stimuli. The response of ON-like LGN neurones to strobe-flash stimulation of the contralateral eye was characterised by a primary excitatory or early discharge (ED) phase after a latency of 25-40 ms. Thereafter, a 200- 400 ms period of inhibition was observed. In 57% of the sample, a rebound excitatory or late discharge (LD) phase completed the response. OFF-like LGN neurones were inhibited by the strobe-flash stimuli after a latency of 30- 35 ms. This flash-evoked inhibition was maintained for 200-400 ms. The sensitivity of the flash-evoked LGN response to the level of background illumination was tested in 11 ON-like and 10 OFF-like neurones. No sustained secondary excitatory events, as observed in visual cortical neurones, were found in any of the ON- and OFF-like LGN neurones, irrespective of the level of background illumination. In conclusion, the data show that the late re-excitatory phase evoked in cortical neurones upon strobe-flash stimulation, is not due to sustained LGN (thalamic) input. Rather, it suggests that these re-excitatory phases are due to intracortical processing of the transient stimuli. These findings emphasize the independent role of the cortex in computing the response to visual stimuli, and cast doubt on traditional theories that have emphasised the role of the thalamus in shaping cortical responses. The difference in the flash and electrically evoked cortical response suggests that even though substantial inhibition is available to the cortex, only a small fraction of this inhibitory capacity is utilised during natural stimulation.

The brain is highly efficient at processing complicated patterns of information, filtering ambiguous input it receives from the senses. Competition between these representations is regulated by multiple mechanisms, together forming a coherent percept of our environment. One approach to regulating incoming sensory information is the recruitment of a canonical neural computation: divisive normalization. Another approach to further steer processing towards behaviorally relevant goals is by means of cognitive influences. In this project I examined the degree to which cognitive processes interact with normalization to shape human visual perception. First, a set of fMRI experiments (Exps 1-3: n=6) examined the hypothesis that attention-driven gain modulation within human visual cortex is dependent on the magnitude of normalization. Leveraging the fact that normalization is modulated by similarity of visual features, my results illustrated that attentional modulation of BOLD responses is larger when visual cortex is put under stronger normalization. These results suggest that the degree to which a subpopulation exhibits normalization plays a role in dictating its potential for attentional benefits. Second, I examined how attention modulates visual population responses (n=7). Neurons within visual cortex exhibit a well-characterized relationship between stimulus intensity and the neural response, known as a contrast response function. While animal and psychophysical studies suggest that attention improves visual processing by multiplicatively increasing the gain of the contrast response, human fMRI studies instead report additive attentional effects, which act independently of stimulus contrast. Consistent with prior work, I demonstrated using a fMRI model-based analysis that attentional modulation indeed appears additive within early visual cortex. Lastly, I examined whether perceptual and memory representations are distinct from one another (Exp 1: n=12, Exp 2: n=10). A prevailing theory posits that the retention of visual memories involves maintenance of information within visual cortices. I tested the degree to which representations in visual memory undergo contrast normalization, by leveraging a classic demonstration: center- surround suppression. I obtained psychophysical measurements of perceived contrast and found robust normalization in perception, yet no signature of normalization occurring between visual memory representations. Taken together, this work helps unravel the underlying neural mechanisms by which cognitive influences shape visual perception.