CoSMo 2018 - Revision history

Gunnar: /* Group Projects */

2018-08-13T17:16:05Z

Group Projects

← Older revision		Revision as of 17:16, 13 August 2018
Line 178:		Line 178:
	'''Project title''': "Hysteresis in stereoscopic fusion: A phenomenon with a surprising amount of depth" <br>		'''Project title''': "Hysteresis in stereoscopic fusion: A phenomenon with a surprising amount of depth" <br>
	'''Abstract''': If each of our eyes sees a line and these are sufficiently close, we can fuse them into a 3D-percept. Research has shown that this phenomenon is history dependent: if we fused signals in the recent past we are more likely to fuse them now. However, whether this hysteresis effect is stimulus-dependent is unknown. Here we show that both a sequential causal inference model and a switching Kalman filter framework predict that hysteresis for fusion/fission scales with a prior for continuity. This prior is built sequentially given stimulus statistics. This suggests that stereo vision is attuned to the statistics of the environment. <br>		'''Abstract''': If each of our eyes sees a line and these are sufficiently close, we can fuse them into a 3D-percept. Research has shown that this phenomenon is history dependent: if we fused signals in the recent past we are more likely to fuse them now. However, whether this hysteresis effect is stimulus-dependent is unknown. Here we show that both a sequential causal inference model and a switching Kalman filter framework predict that hysteresis for fusion/fission scales with a prior for continuity. This prior is built sequentially given stimulus statistics. This suggests that stereo vision is attuned to the statistics of the environment. <br>
			'''WINNER !!!!!''' <br>
	[[Media:Presentation_correct_template_final_POMDP.pdf \| Presentation slides]] and [[Media:Gui_POMDP_CoSMo2018.zip \| GUI for simulations]] <br>		[[Media:Presentation_correct_template_final_POMDP.pdf \| Presentation slides]] and [[Media:Gui_POMDP_CoSMo2018.zip \| GUI for simulations]] <br>

Gunnar: /* Group Projects */

2018-08-13T14:43:38Z

Group Projects

← Older revision		Revision as of 14:43, 13 August 2018
Line 166:		Line 166:
	'''Project title''': "What determines the relationship between motor variability and the speed of motor learning?" <br>		'''Project title''': "What determines the relationship between motor variability and the speed of motor learning?" <br>
	'''Abstract''': Motor variability has been associated with enhanced motor learning, though recent findings question the generality of this effect and the task features that allow this facilitation require clarification. We approximated the optimal learner during a visuomotor learning task (using a Kalman Filter), introducing a perturbation which required learning to maintain optimal performance. We varied features of the model and of the task, measuring the relationship between motor variability and learning for each iteration. Our results show that the relationship between learning and motor variability is determined by the features of the task in question, rather than being a general phenomenon. <br>		'''Abstract''': Motor variability has been associated with enhanced motor learning, though recent findings question the generality of this effect and the task features that allow this facilitation require clarification. We approximated the optimal learner during a visuomotor learning task (using a Kalman Filter), introducing a perturbation which required learning to maintain optimal performance. We varied features of the model and of the task, measuring the relationship between motor variability and learning for each iteration. Our results show that the relationship between learning and motor variability is determined by the features of the task in question, rather than being a general phenomenon. <br>
			[[Media:FinalPresentation_KeepKalm-an_carry_on.pdf \| Presentation slides]] <br>

	''2:30 - 3pm'' \|		''2:30 - 3pm'' \|

Gunnar: /* Group Projects */

2018-08-13T14:19:48Z

Group Projects

← Older revision		Revision as of 14:19, 13 August 2018
Line 209:		Line 209:
	'''Project title''': "Perception of face gender under ambiguity: neural network of face adaptation" <br>		'''Project title''': "Perception of face gender under ambiguity: neural network of face adaptation" <br>
	'''Abstract''': Our perception of the gender of a face is dependent on the recent gender history. Face adaptation studies suggest that constant exposure to one gender shifts the perception to the opposite gender. However, we are currently lacking models that can successfully describe this behavioral phenomenon. We thus implement a neural network model of face gender perception. We find that this model faithfully captures the dependency of face perception on gender history. This demonstrates that perception of gender is influenced by the frequency of exposure to gender faces. <br>		'''Abstract''': Our perception of the gender of a face is dependent on the recent gender history. Face adaptation studies suggest that constant exposure to one gender shifts the perception to the opposite gender. However, we are currently lacking models that can successfully describe this behavioral phenomenon. We thus implement a neural network model of face gender perception. We find that this model faithfully captures the dependency of face perception on gender history. This demonstrates that perception of gender is influenced by the frequency of exposure to gender faces. <br>
			[https://docs.google.com/presentation/d/1jETKox8ah7o53oryDnqIEJMIV2cVMN543o5TN3M8D2c/edit?usp=sharing Presentation slides] <br>

	'''[https://goo.gl/forms/iBRGlpRlZmhYfAYx2 Project voting form]''' (requires Google sign-in to prevent duplicates - no user info recorded, it's an anonymous vote) <br>		'''[https://goo.gl/forms/iBRGlpRlZmhYfAYx2 Project voting form]''' (requires Google sign-in to prevent duplicates - no user info recorded, it's an anonymous vote) <br>

	''6pm +'': '''Closing reception @ Loring Pasta bar''' <br>		''6pm +'': '''Closing reception @ Loring Pasta bar''' <br>

Gunnar: /* Group Projects */

2018-08-13T14:13:17Z

Group Projects

← Older revision		Revision as of 14:13, 13 August 2018
Line 177:		Line 177:
	'''Project title''': "Hysteresis in stereoscopic fusion: A phenomenon with a surprising amount of depth" <br>		'''Project title''': "Hysteresis in stereoscopic fusion: A phenomenon with a surprising amount of depth" <br>
	'''Abstract''': If each of our eyes sees a line and these are sufficiently close, we can fuse them into a 3D-percept. Research has shown that this phenomenon is history dependent: if we fused signals in the recent past we are more likely to fuse them now. However, whether this hysteresis effect is stimulus-dependent is unknown. Here we show that both a sequential causal inference model and a switching Kalman filter framework predict that hysteresis for fusion/fission scales with a prior for continuity. This prior is built sequentially given stimulus statistics. This suggests that stereo vision is attuned to the statistics of the environment. <br>		'''Abstract''': If each of our eyes sees a line and these are sufficiently close, we can fuse them into a 3D-percept. Research has shown that this phenomenon is history dependent: if we fused signals in the recent past we are more likely to fuse them now. However, whether this hysteresis effect is stimulus-dependent is unknown. Here we show that both a sequential causal inference model and a switching Kalman filter framework predict that hysteresis for fusion/fission scales with a prior for continuity. This prior is built sequentially given stimulus statistics. This suggests that stereo vision is attuned to the statistics of the environment. <br>
	[[Media:~~Final_Presentation_no_videos_-_smooth_optimators5~~.pdf \| Presentation slides]] <br>		[[Media:Presentation_correct_template_final_POMDP.pdf \| Presentation slides]] and [[Media:Gui_POMDP_CoSMo2018.zip \| GUI for simulations]] <br>

	''3:30 - 4pm'' \|		''3:30 - 4pm'' \|
Line 183:		Line 183:
	'''Project title''': "Not So Smooth: Modelling Reaching Behaviour of Children with Developmental Coordination Disorder" <br>		'''Project title''': "Not So Smooth: Modelling Reaching Behaviour of Children with Developmental Coordination Disorder" <br>
	'''Abstract''': Reaching movements in healthy subjects usually display one-peak speed profiles. Children with Developmental Coordination Disorder (DCD) instead exhibit intermittent trajectories with multiple peaks in the speed profile. The underlying mechanism is still debated. Here we investigated whether a violation in the internal model of the arm dynamics can cause the observed characteristics of DCD, by means of Optimal Feedback Control (OFC). We found that small changes in the internal model resulted in curved paths and rough speed profiles. We suggest a new hypothesis for DCD deficits: multiple speed peaks could be explained by an inaccurate internal model of arm dynamics. <br>		'''Abstract''': Reaching movements in healthy subjects usually display one-peak speed profiles. Children with Developmental Coordination Disorder (DCD) instead exhibit intermittent trajectories with multiple peaks in the speed profile. The underlying mechanism is still debated. Here we investigated whether a violation in the internal model of the arm dynamics can cause the observed characteristics of DCD, by means of Optimal Feedback Control (OFC). We found that small changes in the internal model resulted in curved paths and rough speed profiles. We suggest a new hypothesis for DCD deficits: multiple speed peaks could be explained by an inaccurate internal model of arm dynamics. <br>
			[[Media:Final_Presentation_no_videos_-_smooth_optimators5.pdf \| Presentation slides]] <br>

	''4 - 4:30pm'' \|		''4 - 4:30pm'' \|

Gunnar: /* Group Projects */

2018-08-13T14:11:14Z

Group Projects

← Older revision		Revision as of 14:11, 13 August 2018
Line 177:		Line 177:
	'''Project title''': "Hysteresis in stereoscopic fusion: A phenomenon with a surprising amount of depth" <br>		'''Project title''': "Hysteresis in stereoscopic fusion: A phenomenon with a surprising amount of depth" <br>
	'''Abstract''': If each of our eyes sees a line and these are sufficiently close, we can fuse them into a 3D-percept. Research has shown that this phenomenon is history dependent: if we fused signals in the recent past we are more likely to fuse them now. However, whether this hysteresis effect is stimulus-dependent is unknown. Here we show that both a sequential causal inference model and a switching Kalman filter framework predict that hysteresis for fusion/fission scales with a prior for continuity. This prior is built sequentially given stimulus statistics. This suggests that stereo vision is attuned to the statistics of the environment. <br>		'''Abstract''': If each of our eyes sees a line and these are sufficiently close, we can fuse them into a 3D-percept. Research has shown that this phenomenon is history dependent: if we fused signals in the recent past we are more likely to fuse them now. However, whether this hysteresis effect is stimulus-dependent is unknown. Here we show that both a sequential causal inference model and a switching Kalman filter framework predict that hysteresis for fusion/fission scales with a prior for continuity. This prior is built sequentially given stimulus statistics. This suggests that stereo vision is attuned to the statistics of the environment. <br>
			[[Media:Final_Presentation_no_videos_-_smooth_optimators5.pdf \| Presentation slides]] <br>

	''3:30 - 4pm'' \|		''3:30 - 4pm'' \|

Gunnar: /* Group Projects */

2018-08-13T14:09:00Z

Group Projects

← Older revision		Revision as of 14:09, 13 August 2018
Line 171:		Line 171:
	'''Project title''': "Modeling feedback gains in a perturbed reaching task" <br>		'''Project title''': "Modeling feedback gains in a perturbed reaching task" <br>
	'''Abstract''': In perturbed reaching tasks, the visuomotor reflex reaction strength has been shown to decrease near the end of movement. However, it remains unclear whether this decrease can be explained by system delays or distance-to-target. Here we use an optimal feedback controller to model reaching movements towards a target, perturbed at different positions. Our results show that system delays, rather than distance-to-target, modulate the visuomotor reflex reaction strength. This suggests that the reflexive reaction strength may be dependent on movement kinematics rather than being pre-set for a given movement. <br>		'''Abstract''': In perturbed reaching tasks, the visuomotor reflex reaction strength has been shown to decrease near the end of movement. However, it remains unclear whether this decrease can be explained by system delays or distance-to-target. Here we use an optimal feedback controller to model reaching movements towards a target, perturbed at different positions. Our results show that system delays, rather than distance-to-target, modulate the visuomotor reflex reaction strength. This suggests that the reflexive reaction strength may be dependent on movement kinematics rather than being pre-set for a given movement. <br>
			[[:Media:Modeling_feedback_gains_in_a_perturbed_reaching_task.pdf \| Presentation slides]] <br>

	'' 3 - 3:30pm'' \|		'' 3 - 3:30pm'' \|

Gunnar: /* Group Projects */

2018-08-11T23:02:52Z

Group Projects

← Older revision		Revision as of 23:02, 11 August 2018
Line 200:		Line 200:
	'''Project title''': "Does motor variability influence learning in uncertain situations?" <br>		'''Project title''': "Does motor variability influence learning in uncertain situations?" <br>
	'''Abstract''': We seem to benefit from variability introduced by our own motor system, but experiencing environmental uncertainty reduces our ability to learn. However, it has not yet been investigated how motor variability modulates the effect of environmental uncertainty on visuomotor adaptation. Here we use an adaptive Kalman filter to predict adaptation behavior under different perturbation uncertainties. We show that for higher motor noise and perturbation uncertainty, adaptation rate and final compensation increase, and difference in motor variance decreases. Therefore, this study shows that larger variabilities in movements aid learning under greater uncertainty. <br>		'''Abstract''': We seem to benefit from variability introduced by our own motor system, but experiencing environmental uncertainty reduces our ability to learn. However, it has not yet been investigated how motor variability modulates the effect of environmental uncertainty on visuomotor adaptation. Here we use an adaptive Kalman filter to predict adaptation behavior under different perturbation uncertainties. We show that for higher motor noise and perturbation uncertainty, adaptation rate and final compensation increase, and difference in motor variance decreases. Therefore, this study shows that larger variabilities in movements aid learning under greater uncertainty. <br>
			[[Media:Cosmo2018_BCP.pdf \| Presentation slides]] <br>

	''5:30 - 6pm'' \|		''5:30 - 6pm'' \|

Gunnar: /* Group Projects */

2018-08-11T23:02:00Z

Group Projects

← Older revision		Revision as of 23:02, 11 August 2018
Line 193:		Line 193:
	Andreea Loredanna Cretu \| Harun Karimpur \| Padmapriya Muralidharan \| Sabrina Hansmann-Roth <br>		Andreea Loredanna Cretu \| Harun Karimpur \| Padmapriya Muralidharan \| Sabrina Hansmann-Roth <br>
	'''Project title''': "To slide or not to slide: Deviations of friction estimation in intuitive physics" <br>		'''Project title''': "To slide or not to slide: Deviations of friction estimation in intuitive physics" <br>
	'''Abstract''': At a glance, we perceive whether the ground is icy or safe to walk on with an intuitive understanding about the physical properties of objects. Here, we investigate how people weight object size and friction in their interpretations. Participants passively view a video of a cube on a rotating plane. They are then asked to tilt the plane to the maximum estimated angle at which they expect the object to start sliding. Our results highlight a common misconception of intuitive physics demonstrating a deviance from an ideal observer model which predicts that size does not play a role for friction estimation.		'''Abstract''': At a glance, we perceive whether the ground is icy or safe to walk on with an intuitive understanding about the physical properties of objects. Here, we investigate how people weight object size and friction in their interpretations. Participants passively view a video of a cube on a rotating plane. They are then asked to tilt the plane to the maximum estimated angle at which they expect the object to start sliding. Our results highlight a common misconception of intuitive physics demonstrating a deviance from an ideal observer model which predicts that size does not play a role for friction estimation. <br>
			[[Media:CoaSC_CoSMo2018.pdf \| Presentation slides]] <br>

	''5 - 5:30pm'' \|		''5 - 5:30pm'' \|

Gunnar: /* Group Projects */

2018-08-11T23:00:46Z

Group Projects

← Older revision		Revision as of 23:00, 11 August 2018
Line 187:		Line 187:
	'''Project title''': "Synchronizing to Decide in Small-World Networks" <br>		'''Project title''': "Synchronizing to Decide in Small-World Networks" <br>
	'''Abstract''': Network neuroscience assumes that synchronous activity across the brain reflects coordination between distinct regions. Although a network’s connectivity structure determines its ability to maintain temporal patterns in the presence of noise, it is unclear how this specifically leads to decisions. To explore this, we model value-based decisions where two oscillating networks suppress each other’s capacity to synchronize. We show the emergence of winner-take-all dynamics and demonstrate that small-world structure provides an advantage in the competition for synchrony in the presence of noise. We conclude that decisions could result from connectivity-dependent competition for synchronous activity between regions of the brain. <br>		'''Abstract''': Network neuroscience assumes that synchronous activity across the brain reflects coordination between distinct regions. Although a network’s connectivity structure determines its ability to maintain temporal patterns in the presence of noise, it is unclear how this specifically leads to decisions. To explore this, we model value-based decisions where two oscillating networks suppress each other’s capacity to synchronize. We show the emergence of winner-take-all dynamics and demonstrate that small-world structure provides an advantage in the competition for synchrony in the presence of noise. We conclude that decisions could result from connectivity-dependent competition for synchronous activity between regions of the brain. <br>
			[[Media:SWoNs_presentation_Combined_CoSMo2018.pdf \| Presentation slides]] <br>

	''4:30 - 5pm'' \|		''4:30 - 5pm'' \|

Gunnar: /* Group Projects */

2018-08-11T19:56:06Z

Group Projects

← Older revision		Revision as of 19:56, 11 August 2018
Line 160:		Line 160:
	'''Project title''': "Modeling Error Source Attribution During Motor Adaptation" <br>		'''Project title''': "Modeling Error Source Attribution During Motor Adaptation" <br>
	'''Abstract''': Humans can maintain accurate motor performance in the face of motor errors, regardless of whether the errors originate internally from their own body or externally from the world. Currently, little is known about how the brain identifies the errors’ source and adapts control accordingly. We designed a linear regressor to discriminate the source of errors in a simulated force field adaptation task. The model is capable of updating body and world-centric parameters driven by internally- and externally-attributed errors, respectively. This model suggests a possible mechanism for attributing a motor error to an internal or external source. <br>		'''Abstract''': Humans can maintain accurate motor performance in the face of motor errors, regardless of whether the errors originate internally from their own body or externally from the world. Currently, little is known about how the brain identifies the errors’ source and adapts control accordingly. We designed a linear regressor to discriminate the source of errors in a simulated force field adaptation task. The model is capable of updating body and world-centric parameters driven by internally- and externally-attributed errors, respectively. This model suggests a possible mechanism for attributing a motor error to an internal or external source. <br>
			[[Media:CoSMo2018_EAGEr_BEAVeRS.pdf \| Presentation slides]] <br>

	''2 - 2:30pm'' \|		''2 - 2:30pm'' \|