Social Networks and Cognitive Biases
Advising Opportunities
School-Specific Advising
Students in Computer Science or Engineering belong to the School of Arts and Sciences or College of Engineering.
Students can receive advice from advisors in those schools.
Dedicated Advising Sessions
The School of Communication and Information has a dedicated meeting scheduled for this Friday during the student session.
Advisors will be present during this meeting.
A reminder link for the advising session can be found on campus resources.
Cognitive Biases in Social Interaction
Definitions of Cognitive Biases
Recency Bias: Refers to the tendency to think more about people you have spoken to most recently. This bias influences recall during conversations.
Salience (Cognitive Salience): Describes how concepts, topics, or relationships that feel more relevant or “safe” come to mind first, affecting recall in social interactions.
Impact on Memory Recall
Both biases impact which names or people individuals recall when queried about recent conversations.
They are considered mundane biases and are distinct from more serious accusations or judgments in social settings.
Social Networks & Connections
Definitions and Structure
Social Network: A collection of nodes (representing people, devices, documents) connected by edges (defining relationships, hyperlinks, interactions).
In a graph model:
Each node stands for an entity (e.g., a person, a computer, a document).
Edges symbolize relationships (e.g., friendship, colleague, hyperlink).
Edge types can be encoded as labels.
Network Metrics & Graph Concepts
Overview of Metrics
Network Metric: A numerical value derived from the quantity and quality of links between nodes; used to infer properties of the network.
Procedure to Analyze a Network
Identify all nodes in the system.
Define the relationship type for edges (e.g., friendship, citation).
Count the number of edges each node has (degree).
Weight edges by quality (e.g., strength of relationship, frequency of interaction).
Compute metrics (e.g., weighted degree, centrality) to assess the importance of nodes.
Platforms Comparison (LinkedIn vs. Facebook)
Insights from Class Discussion
The comparisons are based on self-reported estimates from class discussion; they are not derived from a formal study.
Relationship with Platform Types
LinkedIn: Primarily used for professional contacts.
Few users feel comfortable asking many contacts for favors; often limited to a small subset trusted.
Facebook: Used for social and friendship connections.
Perception of “real friends” varies; many consider Facebook friends to be less reliable for financial requests.
Research Findings on Conversation Counts
Key Findings
Students were asked to list all people they had spoken with, with a typical range of 15–20 contacts before they start to forget or give up counting.
This aligns with research indicating a cognitive limit on maintaining active conversational ties.
Analyzing a Social Network (Step-by-Step Procedure)
Data Collection and Graph Creation
Collect data on who spoke to whom (e.g., conversation logs, self-reports).
Create a graph representation where each individual is a node.
Add edges for each conversation and label them with the relationship type if relevant.
Calculation of Basic Metrics
Degree: Number of connections per node.
Weighted Degree: Used when edge strengths are known.
Interpretation of Results:
Analyze ability to borrow $20, integrating platform types influence:
LinkedIn: Professional contacts have less trust for financial requests.
Facebook: Social/friend connections are also less trustworthy for similar requests.
High-Degree Nodes and Social Bonds
Characteristics of Nodes
High-degree nodes may act as hubs or central figures in the network.
Edge labels reveal the types of social bonds (e.g., friendship, classmate, advisor).
These insights can be applied to real-world contexts like advising outreach and networking strategies.
Key Takeaways
Be aware of specific advising times within your school schedule.
Recognize that recency bias and salience influence your memory of conversations.
A social network consists of a graph formed by nodes (people/devices) and edges (relationships).
Network metrics help quantify the importance of nodes based on the number and quality of links.
Differences between platforms (LinkedIn vs. Facebook) influence trust levels regarding requests.
There is a cognitive limit, as people generally can manage 15–20 active conversational connections.
Network Basics
Basic Definitions
Network Model: An abstraction representing entities (nodes) and their relationships (edges).
Node: A noun representing an individual element (e.g., a student, an actor).
“Every network has three things: a noun, another noun (which could be a node), and a connection between both.”
Edge: A line indicating a specific type of relationship (e.g., friendship, classmate).
Social Network Modeling
Social circles can be visualized as graphs based on friend connections:
Friends → 1-hop connections.
Friends of friends → 2-hop connections.
Higher-order acquaintances → further connections.
Mutual Friend: A node connecting two otherwise separate sub-graphs.
Bridge (Articulation Point): A node whose removal splits the network into disconnected components.
Example scenario: “When we remove Johnette, we turn one network into two.”
Measuring Popularity
Steps to Compute Degree Centrality
List all edges incident to the node.
Count the edges (or sum their weights if weighted).
Rank nodes based on the resulting count.
Path & Hop Concepts
Definitions and Procedure
Trace Route: Refers to finding all possible paths between two nodes while counting hops.
k-hop Neighbor: Any node reachable in exactly k edges.
Example Quote: “Run a trace route across our entire friendship network” to discover 1-hop, 2-hop, … 5-hop relationships.
Node Importance & Removal
Articulation Point Detection
Identifying nodes whose removal increases the number of connected components will directly affect the network’s structure and functionality.
Impact of Removal:
Loss of connectivity between sub-networks.
A decrease in overall network cohesion.
Example: “When we remove this person who connects two networks, we refer to them as bridging between two sets of practice.”
Key Network Metrics and Examples
Degree Centrality: The count of direct edges indicating how many people sat with you.
Example: “John N had a lot of buddies at the table.”
Betweenness Centrality: Frequency with which a node lies on the shortest paths between others, indicating its bridge role.
Example: “Johnette is between the East Coast and West Coast groups.”
Weighted Degree: Represents edge weight equivalent to the number of reported interactions.
Example: “Circle size represents the number of users who reported sitting together.”
Hollywood Actor Network Analysis
Network Representation
Nodes: Actors (for example, Samuel L. Jackson).
Edges: Co-appearance in the same movie provides the basis for connections.
Application of Analysis Tools
Tools used for the lunch-room network also apply to this actor network analysis:
Degree Centrality: Identifies the most collaborative actors.
Betweenness Centrality: Identifies actors who link otherwise separate film clusters.
Example Visualization (JEPI / DEPTHI)
Graphical Representation
Nodes are displayed as circles with size proportional to interaction counts. Edges shown as lines, each line equals a reported sitting-together event.
Visual cues highlight:
Large nodes indicating high popularity.
Bridging nodes critical for network connectivity.
Core Graph Concepts (with LaTeX Formulas)
Key Graph Formulas
Degree of a Node:
ext{deg}(v)
Shortest Path Length Between Nodes:
d(s,t) = |p|
where $p$ is the set of all paths.Betweenness Centrality:
CB(v) = rac{ ext{Total number of shortest paths passing through } v}{ ext{Total number of shortest paths}}
Applications of Graph Concepts
Use these concepts to model any social setting (e.g., high-school lunch tables, online friend circles, or Hollywood collaborations) and apply graph algorithms to uncover hidden structures and influential participants.
Network Centrality & Node Removal
Importance of Central Nodes
Central nodes act as primary bridges linking various parts of the network.
If a central node (e.g., a founder, lead engineer, or a key individual) leaves, sub-networks risk becoming disconnected, leading to organizational disintegration.
The disconnection proves critical, as many network members may be connected solely through that single central communication line.
Implication of Network Centrality
Definition: Network centrality measures a node’s importance based on its position within the network, often reflecting how many other nodes rely on it for connection.
Key Mathematical Expressions for Centrality
deg(v) = ext{Count of direct edges}
d(s,t) = ext{length of shortest path}
rac{|Ns||Nt|}{|E|} = rac{|v| imes|u|}{|E|}
Disconnection Matters
Degrees of Separation
1st-degree connections: Direct links (e.g., “I worked with Sam Jackson”).
2nd-degree connections: Friends of friends (can ask a mutual contact for assistance).
Beyond 2nd degree: These become acquaintances, where influence diminishes sharply.
### Example Path Extension (Viggo Mortensen → Sam Jackson)
Viggo Mortensen → Lord of the Rings → Actor A
Actor A → Terminator → Actor B
Actor B → Titanic → Leonardo DiCaprio
Leonardo DiCaprio → Inception → Actor C
Actor C → Sam Jackson (through a movie)
Degree Connection
This example illustrates a 5-degree connection (friends of friends, extending up to five steps).
Network Analysis in Organizations
Data Collection Steps
Companies collect data to understand communication dynamics and collaboration patterns:
Collect basic interaction data (e.g., emails, meetings, project logs).
Determine consequences based on analysis:
Loss of Communication: Loss of exchange of information if a central hub is removed.
Reduced Collaboration: Projects stalling due to lack of input from diverse teams.
Organizational Fragility: Vulnerability of the entire system when one individual leaves.
Steps for Data-Driven Network Insight
Analyze the structure to identify:
Central nodes,
Bottlenecks,
Clusters.
Visualize the network for stakeholders to gain actionable insights.
Make decisions regarding:
Redesigning communication pathways,
Redistributing responsibilities,
Reinforcing vulnerable network links.
Highlight the role of a graduate in all four stages of data collection, analysis, visualization, and decision-making.
Hollywood Actor Network Using IMDb
Building the Actor Network Graph
Using a movie database such as IMDb and tools like Oracle Lincoln, we can visualize actors as nodes who co-appear in films, defining edges based on collaborative appearances with:
Direct connections with Sam Jackson (1-degree),
Connections of those actors with others (2-degree) etc.
Observational Insights
The “Kevin Bacon” Game establishes the shortest path length between any actor and Kevin Bacon using co-appearance networks.
This originated from a college trivia game noting Bacon’s frequent film appearances.
It illustrates small-world properties where most actors are only a few steps away from one another.
Fact of Interest
The “six degrees of Kevin Bacon” concept resonates with the broader six-degree-of-separation phenomenon existing in social networks.
Analysis Observations
Top 20 Central Actors: Dominate the network, indicating a concentration of influence.
Gender Representation: No actresses appear at the very core of the central nodes (based on latest data), suggesting systemic representation issues in Hollywood.
Racial Diversity: Limited representation among central nodes indicates further systemic concerns.
Key Terms
Centrality: The importance level of a node based on its position within the network.
Degree (of Separation): The total number of edges in the shortest path between two nodes.
Network Fragility: The susceptibility of a network to collapse upon the removal of its central nodes.
Small-World Network: A network structure where most nodes can be reached from any others in a few steps.
Graph Visualization: The process of creating graphical representations of nodes and edges for analytical purposes.