Recitation

• Attention


  The mental process of focusing on one aspect amidst multiple potential stimuli
• Attention


  • Voluntary: meaning we can choose what to focus on within our environment
  • Selective: Involves prioritizing some thoughts or objects over others
  • Limited: indicating that it cannot be evenly distributed across many objects simultaneously
  • It typically allows us to perform only one physical action at a time
• Effective attention


  Requires withdrawing from certain stimuli to manage others more efficiently, contrasting with a state of confusion or distraction
• Attention


  • Aids in simplifying brain processing by filtering out irrelevant signals and boosting the reliability of relevant ones
  • Plays a crucial role in working memory and awareness, enabling focus on both external and internal representations, such as specific memories
• Saccade


  A rapid movement of the eye between fixation points
• Overt attention


  Directing eyes towards a point to process information – saccade on target area
• Covert attention


  • Focus without eye movement – neural selection without saccade
  • Enhances the brain's early visual response, allowing us to better process visual information without the need for direct eye movement
• Attention


  • Concentrates on relevant stimuli, filters out distractions, enhances visual responses
  • Enhances neural signals similarly to increasing a stimulus's physical intensity
  • Can focus on specific features, like color or motion, regardless of spatial location
  • Increases the brain's response to stimuli and employs inhibitory mechanisms to suppress distractions
  • Object-focused attention enhances processing of features like color or motion, regardless of location
• Superior colliculus (SC)


  • Critical for eye movement initiation; must reach an activation threshold to trigger saccades
  • Its indirect measurement linked to cortical areas like the parietal cortex and FEF, involved in starting saccades
• Prefrontal cortex (PFC)


  Influences visual area responses; modulates attention
• Frontal eye fields (FEF)


  Involved in inhibiting and inducing saccades, indicating a role in both covert and overt attention
• Treisman's Feature Integration Theory (FIT)


  • An individual combines different visual features to perceive a complete object
  • Preattentive Stage: The visual system processes various features of objects in the environment simultaneously and in parallel, without the need for focused attention
  • Focused Attention Stage: Attention is required to bind these independently processed features together at a specific location to form a unified coherent perception of the object
• Attention Modulation Index (AMI)


  • (attended - unattended) / (attended + unattended)
  • AMI > 0 enhancement; AMI < 0 suppression
• Brain imaging was used to visualize the areas in the LGN where attention either amplifies or suppressed neural activity based on the participants' focus
• Attentional modulation in the LGN and SC, correlated with specific locations in the visual field
• Various streams of visual processing, which originate from different types of retinal ganglion cells and travel separately through the subcortical areas before converging in the visual cortex, are differentially affected by cognitive control mechanisms
• Attention can either enhance or suppress the neural response to visual stimuli in both LGN and SC, these effects can be related to the location of the stimuli in the visual field (retinotopy)
• Dorsal Attention Network (DAN)


  • Active when we choose where to focus based on our own goals
  • Involved when we pay attention to particular features, objects, or places
  • Shows continuous activity during tasks that need extended attention
  • Activity levels can indicate potential performance on attention-driven tasks
  • Includes the intraparietal sulcus and frontal eye fields, important for covert attention and panning movements
  • Surrounding parietal areas manage control processes like response planning and task management
• Ventral Attention Network (VAN)

  • Consists of the temporoparietal junction (TPJ) and ventral frontal cortex (VFC)
  • Specialized in responding to unexpected but important events
  • Responds to unexpected stimuli like surprises or novel events, primarily in the right hemisphere's temporoparietal and inferior frontal areas
  • Activates strongly when reality does not match expectations
  • Engages in stimulus-driven attentional shifts, particularly for significant, goal-related stimuli
  • The TPJ, a component of the VAN, is finely tuned to detect behaviorally relevant stimuli and directs attention accordingly
  • May act as a "reset signal" to prompt shifts in behavior and attention in alignment with new objectives or demands
• Bottom-up Attention (stimulus driven)

  • It is feature-driven, meaning it is automatically captured by the salient features of the environment, such as bright colors or loud sounds
  • Knows as reflexive or automatic, this type of attention does not require conscious effort
  • Termed exogenous as it is externally guided by the stimuli
  • It operates quickly, resting fast to changes in the environment
• Top-down Attention (goal directed)

  • Driven by goals or past experiences, implying it is under voluntary control based on what we deem important at the moment
  • Considered endogenous because it comes from internal decisions about where to focus attention
  • It operates at a slower pace as it involves conscious direction and more often deliberate shifting of attention
• Working Memory


  A cognitive system with limited capacity, responsible for temporarily holding information available for processing
• Working Memory


  • Maintenance (Storage and Rehearsal): Keeping information in working memory through active processes, such as repeating a phone number until you can save it
  • Representations: Symbolic codes for information that are stored in neuronal networks, either transiently or permanently
  • Control (Elaboration): Managing the information in working memory, such as using an inner voice for rehearsal
  • Operations: The processes or computations that are performed on the representations within working memory
• Working memory is working (control) with your short-term memory representations
• Verbal Working Memory (VWM)


  • Responsible for temporarily storing verbalizable information, such as letters, worlds, numbers, or namable objects
  • Phonological loop: a phonological store that serves to temporarily hold verbal information
• Visuospatial Working Memory (VWM)

  • The ability to retain and process an object's identity and spatial location is essential for many daily tasks
  • Maintenance of spatial information, including the location, shapes, and colors, in short-term memory
  • Involve the manipulation of visual representations, such as remembering the sequence of events, mentally navigating a space, or performing mathematical calculations in your head
  • Visuospatial Sketchpad: ability temporarily to hold visual and spatial information, such as the location of a parked car, or the route from home to a grocery store
• Prefrontal Cortex (PFC)


  • Jacobsen's experiments demonstrated the PFC's role in "immediate memory" by showing impaired performance in monkeys on delayed-response tasks following PFC lesions
  • Malmo's study challenged this by showing performance improvement in the dark, suggesting PFC lesions affected the use of STM amid competing stimuli, not STM itself
  • These findings contributed to the evolution of the concept of "working memory," emphasizing the PFC's role in managing attention and resisting interference, beyond mere STM
  • Neuron in monkey PFC exhibits increased firing rates during an oculomotor delayed-response task, indicating a role in spatial working memory, showing specialization for holding spatial information
  • Suggests organization of the visual working memory system in the brain, with pathways from visual processing areas projecting into specific PFC areas
• Spatial Working Memory


  dlPFC near the principal sulcus involved in spatial memory ("where" memory), responsible for tracking object locations
• Object Working Memory


  Ventral PFC specializes in "what" memory, handling object identities like color and form
• Macaque Studies: Neural activity in the dlPFC is crucial for memory-guided saccades
• Effects of Lesion: decline in WM performance in monkeys with dlPFC lesions, particularly affection the memory for saccade direction in a contralateral (opposite) visual field, as evidenced by Funahashi et al.'s research
• Human studies: damages/disturbance to PCS significantly affect spatial working memory
• Lesions led to memory errors increasing for saccades to the visual field opposite the lesions
• TMS Perturbation studies: TMS to PCS, not dlPFC, during memory tasks increases memory errors. Indicates PCS's role in maintaining spatial information
• The Basic Processing Element

  • Each node or neuron in the network functions by receiving inputs (IN), which are then weighted (WE) by their receptive important or strength
  • The inputs are multiplied by these weights and summed up to form a total weighted input (Σg)
  • This sum is then passed through an activation function (AF), which determines the neuron's output based on the summed input
  • The activation function decides how to transform the input into an output, like a threshold that needs to be crossed for the neuron to activate
  • The output of each neuron can serve as an input to the neurons in the following layer, creating a complex network of signals
• Feed-Forward Networks


  • Neurons process data sequentially from input to output
  • No feedback loops
  • Each neurons sums weighted inputs and applies an activation function
  • Final output is a vector, with each part representing an output neuron's result
  • Networks map inputs to outputs, adjusting weights via learning algorithms
  • Error-backpropagation adjusts weights to minimize the difference between actual and desired outputs
• Recurrent Neural Networks (RNNs)

  • Have feedback loops, recognizing sequential dependencies with an intrinsic 'memory'
  • They constantly update, adapting to new data
  • Trained with Hebbian rules, RNNs can retrieve patterns from noisy inputs
  • "Cells that fire together, wire together."
  • If nodes (neurons) are activated at the same time, the connection between them is strengthened
  • Stabilization: through iterations, they reach a stable state or point attractor
  • Point attractors: Represent stored patterns, enabling precise recall from partial data
  • Serve as ANNs: their adaptability and predictive capabilities in handling sequences
• Point Attractor Neural Networks (PANNs)

  • Have discrete stable states or point attractors
  • Used for associative memory; can recall patterns from partial inputs
  • Evolve towards the nearest attractor based on initial conditions
• Continuous Attractor Neural Networks (CANNs)


  • Feature as continuous manifold of stable states
  • Act as short-term memory, maintaining information over time
  • Suitable for tasks requiring sustained precision
  • Not ideal for discrete pattern storage due to sensitivity to perturbations
• Bistability


  • Network architecture: these cells can also excite a group of interneurons, which in turn provide feedback inhibition to the pyramidal cells
  • The interaction between excitatory and inhibitory connections bistable system
  • State space dynamics: firing rates of interneurons and pyramidal cells interact
  • Trajectory and bistability: excitatory input can push the network into a high firing rate state, which remains after the stimulus stops; inhibitory input can revert it to a low firing rate state
  • Simulation example: bistable behavior with persistent high activity in response to stimulation
  • Different receptoras (NMDA, GABA) neuromodulators (e.g. DA) can influence the stability and range of these states, affecting how easily the network can switch between them