76 research outputs found
Non-linear integration of crowded orientation signals
No full textAbstractCrowding of oriented signals has been explained as linear, compulsory averaging of the signals from target and flankers [Parkes, L., Lund, J., Angelucci, A., Solomon, J. A., & Morgan, M. (2001). Compulsory averaging of crowded orientation signals in human vision. Nature Neuroscience, 4(7), 739–744]. On the other hand, a comparable search task with sparse stimuli is well modeled by a ‘Signed–Max’ rule that integrates non-linearly local tilt estimates [Baldassi, S., & Verghese, P. (2002). Comparing integration rules in visual search. Journal of Vision, 2(8), 559–570], as reflected by the bimodality of the distributions of reported tilts in a magnitude matching task [Baldassi, S., Megna, N., & Burr, D. C. (2006). Visual clutter causes high-magnitude errors. PLoS Biology, 4(3), e56]. This study compares the two models in the context of crowding by using a magnitude matching task, to measure distributions of perceived target angles and a localization task, to probe the degree of access to local information. Response distributions were bimodal, implying uncertainty, only in the presence of abutting flankers. Localization of the target is relatively preserved but it quantitatively falls in between the predictions of the two models, possibly suggesting local averaging followed by a max operation. This challenges the notion of global averaging and suggests some conscious access to local orientation estimates
Reward sharpens orientation coding independently on attention
No full textRewarding improves performance. Is it due to modulations of the output modules of the neural systems or are there mechanisms favoring more 'generous' inputs? Some recent study included V1 in the the circuitry of reward-based modulations, but the effects of reward can easily be confused with effects of attention. Here we address this issue with a psychophysical dual task to control attention while orientation sensitivity on targets associated to different levels of reward is measured. We found that different reward rates improve orientation discrimination and sharpen the internal response distributions. Data are unaffected by changing attentional load nor by dissociating the feature of the reward cue from the feature relevant for the task. This suggests that reward may act independently on attention by modulating the activity of early sensory stages, perhaps V1, through a SNR improvement of task-relevant channels. Reward acts like attention, but using separate channels
Rootstock and Crop Load Effects on ‘Honeycrisp’ Photosynthetic Performance and Carbohydrate Accumulation
No full textRootstock selection and crop load adjustment are key practices in apple orchard management; nevertheless, the effects of rootstocks and crop load levels on important physiological processes of the scions, such as photosynthetic performance and carbohydrate accumulation, are still unclear. To investigate the impact of different rootstocks and crop load levels on scion photosynthesis and carbohydrate buildup, in 2020, ‘Honeycrisp’ trees grafted on rootstocks ‘G.41’, ‘G.935’, and ‘M.9-T337’ were thinned to low and high crop load levels, and photosynthetic performance and carbohydrate accumulation in leaves and fruit were evaluated. Leaves from ‘G.935’ showed the highest net photosynthesis and electron use efficiency of photosynthesis and the lowest activity for non-net carboxylative processes, all together indicative of enhanced photosynthetic performance. High crop load determined an increase in gas exchange, suggesting a positive feedback of high fruit competition on carbon assimilation. While rootstock ‘M.9-T337’ showed a higher accumulation of starch in leaves, no pattern regarding the composition of leaf-soluble sugars among rootstocks could be identified. Conversely, by the end of the harvest season, leaves from low-cropping trees had higher fructose, glucose, and sorbitol than those from high-cropping trees, but differences in starch content were not significant. Fructose and sorbitol concentrations were affected by rootstock and crop load, respectively. Overall, this study showed that high cropping enhanced photosynthesis in ‘Honeycrisp’ apple and determined lower accumulation of some soluble carbohydrates (fructose, glucose, sorbitol) in leaves. This study also provided insights into how rootstocks affect photosynthetic performance of ‘Honeycrisp’, highlighting ‘G.935’ as the rootstock conferring the highest photosynthetic capacity under the present experimental conditions
Differences in time course activation of dorsolateral prefrontal cortex associated with low or high risk choicesin a gambling task
No full textPrefrontal cortex plays an important role in decision making (DM), supporting choices in the ordinary uncertainty of everyday life. To assess DM in an unpredictable situation, a playing card task, such as the Iowa Gambling Task (IGT), has been proposed. This task is supposed to specifically test emotion-based learning, linked to the integrity of the ventromedial prefrontal cortex (VMPFC). However, the dorsolateral prefrontal cortex (DLPFC) has demonstrated a role in IGT performance too. Our aim was to study, by multichannel near-infrared spectroscopy, the contribution of DLPFC to the IGT execution over time. We tested the hypothesis that low and high risk choices would differentially activate DLPFC, as IGT execution progressed. We enrolled 11 healthy adults. To identify DLPFC activation associated with IGT choices, we compared regional differences in oxy-haemoglobin variation, from baseline to the event. The time course of task execution was divided in four periods, each one consisting of 25 choices, and DLPFC activation was distinctly analyzed for low and high risk choices in each period. We found different time courses in DLPFC activation, associated with low or high risk choices. During the first period, a significant DLPFC activation emerged with low risk choices, whereas, during the second period, we found a cortical activation with high risk choices. Then, DLPFC activation decreased to non-significant levels during the third and fourth period. This study shows that DLPFC involvement in IGT execution is differentiated over time and according to choice risk level. DLPFC is activated only in the first half of the task, earlier by low risk and later by high risk choices. We speculate that DLPFC may sustain initial and more cognitive functions, such as attention shifting and response inhibition. The lack of DLPFC activation, as the task progresses, may be due to VMPFC activation, not detectable by fNIRS, which takes over the IGT execution in its second half
Entropy-SGD: biasing gradient descent into wide valleys
No full textThis paper proposes a new optimization algorithm called Entropy-SGD for training deep neural networks that is motivated by the local geometry of the energy landscape. Local extrema with low generalization error have a large proportion of almost-zero eigenvalues in the Hessian with very few positive or negative eigenvalues. We leverage upon this observation to construct a local-entropy-based objective function that favors well-generalizable solutions lying in large flat regions of the energy landscape, while avoiding poorly-generalizable solutions located in the sharp valleys. Conceptually, our algorithm resembles two nested loops of SGD where we use Langevin dynamics in the inner loop to compute the gradient of the local entropy before each update of the weights. We show that the new objective has a smoother energy landscape and show improved generalization over SGD using uniform stability, under certain assumptions. Our experiments on convolutional and recurrent networks demonstrate that Entropy-SGD compares favorably to state-of-the-art techniques in terms of generalization error and training time
- …
