UNIT 3 PROBABILISTIC PROGRAMMING LANGUAGE

Here's an expanded version of the notes on WebPPL syntax with examples:

!) WebPPL: Syntax

Introduction to WebPPL Syntax:

WebPPL, standing for Web Probabilistic Programming Language, serves as a versatile tool for
expressing probabilistic models and conducting inference tasks. Its syntax, reminiscent of
JavaScript, offers a straightforward yet powerful means to model complex probabilistic
systems.

Variables and Constants:

In WebPPL, variables are used to store values, while constants represent fixed values. Variables
are declared using the `var` keyword, and constants can be numerical, boolean, or strings. This
flexibility enables users to define and manipulate data within their models.

SYNTAX:

var x = 5; // Variable declaration

var y = true;

var name = "WebPPL"; // String constant

Probability Distributions:
WebPPL provides a rich set of built-in probability distributions to model uncertainty. These
distributions encompass both discrete and continuous distributions, allowing users to capture
various probabilistic phenomena. Functions such as `flip` for Bernoulli distributions and
`uniform` for continuous uniform distributions facilitate sampling from these distributions.

SYNTAX:

var coinFlip = flip(0.5); // Bernoulli distribution

var uniformSample = uniform(0, 1); // Continuous uniform distribution

Conditionals and Loops:
Conditionals and loops play a vital role in controlling the flow of execution within WebPPL
programs. The `if` statement enables branching based on logical conditions, while loops, such
as the `for` loop, facilitate repetitive execution of code blocks. This control flow mechanism
enhances the expressiveness of WebPPL programs, enabling complex decision-making and
iterative processes.

SYNTAX:

if (coinFlip) {

console.log("Heads");

} else {

console.log("Tails");

SYNTAX:

for (var i = 0; i < 5; i++) {

console.log(i);

Functions:
Functions in WebPPL encapsulate reusable blocks of code and facilitate modular programming.
They can take parameters and return values, allowing for abstraction and organization of code.
With functions, users can create clear and concise representations of probabilistic processes,
promoting code reuse and maintainability.

SYNTAX:

function double(x) {

return x * 2;

}
Probabilistic Models:
In WebPPL, probabilistic models are constructed by combining variables, probability
distributions, and conditioning. Users can define prior distributions to express uncertainty
about model parameters and condition on observed data to update beliefs. This modeling
approach enables the representation of complex probabilistic relationships and facilitates
inference tasks.

SYNTAX:

var x = uniform(0, 10); // Prior distribution

var observedData = 5;

condition(x > observedData); // Conditioning on observed data

Inference Algorithms:
WebPPL offers a range of built-in inference algorithms to perform probabilistic inference. These
algorithms, such as Markov Chain Monte Carlo (MCMC) and Maximum A Posteriori (MAP)
estimation, enable users to infer latent variables and make predictions from observed data. By
leveraging these algorithms, users can extract meaningful insights from probabilistic models
and make informed decisions.

SYNTAX:

var posterior = Infer({method: 'MCMC', samples: 1000}, function() {

var x = uniform(0, 10);

condition(x > observedData);

return x;

});

Conclusion:
Understanding the syntax of WebPPL is essential for effectively expressing probabilistic models
and conducting inference tasks. By mastering the syntax elements discussed above, users can
harness the full capabilities of WebPPL to model complex probabilistic systems, perform
inference tasks, and gain valuable insights from probabilistic data. With its intuitive syntax and
powerful features, WebPPL serves as a versatile tool for probabilistic programming across
various domains.
2) Manipulating Probability Types and Distributions in Cognitive Science

Introduction:
Probability theory serves as a foundational framework in cognitive science, providing a
mathematical basis for understanding and modeling various cognitive processes. Manipulating
probability types and distributions allows researchers to represent uncertainty, make
predictions, and gain insights into human cognition.

Probability Types:

1. Frequentist Probability:

-Definition: Frequentist probability is based on the relative frequency of an event occurring in

repeated trials.

- Explanation: In frequentist probability, the probability of an event is interpreted as the long-

run proportion of times the event occurs in a large number of trials under identical conditions.

- Example: Rolling a fair six-sided die and observing the frequency of each face over a large
number of rolls represents frequentist probability.

2. Bayesian Probability:

- Definition: Bayesian probability represents degrees of belief or uncertainty about events,

incorporating prior knowledge and new evidence.

- Explanation: Bayesian probability allows for the updating of beliefs in light of new evidence
using Bayes' theorem, which combines prior probabilities with likelihoods to compute posterior
probabilities.

- Example: Bayesian inference is commonly used in cognitive science to model learning and
decision-making processes, where prior beliefs are updated based on observed data.

3. Subjective Probability:

- Definition: Subjective probability reflects an individual's personal assessment of the

likelihood of an event, often influenced by cognitive biases and heuristics.

- Explanation: Subjective probability is based on an individual's subjective judgment or

intuition about the likelihood of an event occurring, which may differ from objective or statistical
probabilities.

- Example: A person's estimate of the probability of winning a game of chance may be

influenced by their level of confidence, past experiences, and emotional state.
Probability Distributions:

1. Normal Distribution:

- Definition: The normal distribution describes continuous random variables with a bell-
shaped curve, characterized by a mean and standard deviation.

- Explanation: The normal distribution is symmetrical around its mean, with the majority of
data points clustered near the center and fewer data points in the tails.

- Example: Human characteristics such as height and intelligence often follow a normal
distribution in the population.

2. Binomial Distribution:

- Definition: The binomial distribution represents the probability of a binary outcome (success
or failure) in a fixed number of independent trials.

- Explanation: The binomial distribution is characterized by two parameters: the probability of

success in each trial (p) and the number of trials (n).

- Example: Flipping a fair coin multiple times and counting the number of heads follows a
binomial distribution.

3. Poisson Distribution:

- Definition: The Poisson distribution models the number of events occurring in a fixed interval
of time or space, given the average rate of occurrence.

- Explanation: The Poisson distribution is characterized by a single parameter, lambda (λ),

representing the average rate of events.

- Example: The number of phone calls received by a call center in an hour may follow a Poisson
distribution.

4. Exponential Distribution:

- Definition: The exponential distribution characterizes the time between events in a Poisson
process, representing the waiting time until the next event occurs.

- Explanation: The exponential distribution is characterized by a single parameter, lambda (λ),

representing the rate parameter.

- Example: The time between arrivals of consecutive customers at a service counter may
follow an exponential distribution.
Manipulating Probability Types and Distributions:
- Parameter Estimation: Estimating model parameters from empirical data, such as mean and
variance in a normal distribution.

- Model Comparison: Comparing different probability distributions to evaluate competing

cognitive models.

- Prediction and Inference: Making predictions about future events and performing probabilistic
inference based on observed data.

- Uncertainty Quantification: Quantifying and representing uncertainty in cognitive models to

aid decision-making and risk assessment.

3) Inference in Cognitive Science

Introduction
In cognitive science, inference serves as a cornerstone process, facilitating understanding,
decision-making, and problem-solving in various cognitive tasks. It involves the utilization of
available information, often uncertain or incomplete, to generate educated guesses, draw
conclusions, and make predictions. Understanding the mechanisms underlying inference is
crucial for unraveling the complexities of human cognition across diverse domains.

Types of Inference:

1. Perceptual Inference:

Perceptual inference encompasses the brain's ability to interpret sensory inputs and derive
meaningful representations of the external world. It involves processes such as feature
detection, pattern recognition, and scene segmentation, where the brain integrates sensory
information with prior knowledge to construct coherent perceptual experiences.

2. Probabilistic Inference:

Probabilistic inference deals with reasoning under uncertainty, where the brain evaluates
probabilities to assess the likelihood of different outcomes. It plays a vital role in decision-
making, learning, and prediction, enabling individuals to weigh evidence, estimate risks, and
make optimal choices in ambiguous situations.
3. Causal Inference:

Causal inference involves inferring cause-effect relationships between variables based on

observed data. It allows individuals to identify the underlying mechanisms driving observed
phenomena, discerning causal factors from mere correlations and facilitating effective
intervention strategies.

4. Social Inference:

Social inference pertains to understanding and predicting the thoughts, intentions, and
behaviors of others. It involves mentalizing, perspective-taking, and attributing mental states to
oneself and others, enabling social interactions, empathy, and cooperation.

Methods of Inference:

1. Inductive Reasoning:

Inductive reasoning involves generalizing from specific instances to broader principles or

rules. It allows individuals to extract patterns, infer regularities, and make predictions about
future events based on past experiences.

2. Deductive Reasoning:

Deductive reasoning entails deriving specific conclusions from general principles or

assumptions. It involves logical inference, where valid deductions are made based on
established premises, facilitating problem-solving and logical inference tasks.

3. Abductive Reasoning:

Abductive reasoning entails inferring the best explanation or hypothesis given observed
evidence. It involves generating plausible hypotheses, evaluating their explanatory power, and
selecting the most likely explanation to account for observed phenomena.

Applications of Inference in Cognitive Science:

1. Cognitive Modeling:

Inference techniques are essential for developing computational models of cognitive

processes, simulating how humans perceive, learn, and reason in different contexts. These
models provide insights into the underlying mechanisms of cognition and aid in hypothesis
testing and theory development.
2. Decision-Making and Problem-Solving:

Inference processes underlie decision-making and problem-solving, as individuals evaluate

options, weigh evidence, and select actions based on their beliefs and preferences.
Understanding these processes can inform strategies for improving decision-making
performance and optimizing cognitive performance.

3. Learning and Memory:

Inference mechanisms play a crucial role in learning and memory processes, as individuals
integrate new information with existing knowledge, infer causal relationships, and retrieve
relevant memories to guide behavior. These processes are fundamental for adaptive behavior
and knowledge acquisition.

4. Language Processing:

Inference is integral to language comprehension and production, as individuals infer the

meaning of words and sentences, resolve ambiguities, and make predictions about upcoming
linguistic input. Understanding these processes can shed light on the mechanisms underlying
language understanding and production in the human brain.

Challenges and Future Directions:

1. Neurocomputational Modeling:

Advancing neurocomputational models that integrate neural and cognitive mechanisms to

elucidate the neural basis of inference processes in the brain.

2. Interdisciplinary Collaboration:

Promoting interdisciplinary collaboration between cognitive scientists, neuroscientists,

computer scientists, and statisticians to develop comprehensive models of inference that
capture its multifaceted nature across different levels of analysis.

3. Technological Innovations:

Harnessing technological innovations such as artificial intelligence, machine learning, and

computational modeling to develop more sophisticated and realistic models of human
inference.
Conclusion:

Inference is a fundamental cognitive process that underlies perception, learning, decision-

making, and social interaction in humans. By studying inference mechanisms in cognitive
science, researchers can gain insights into the computational principles that govern human
cognition and develop more accurate models of how the mind works. Continued research in
this area holds the promise of unraveling the mysteries of human thought and behavior, paving
the way for transformative advancements in cognitive science and beyond.

4) Exploring Random Computation

Introduction:

Random computation serves as a fundamental aspect of cognitive science, offering a lens

through which researchers examine the intricate interplay between randomness and human
cognition. This exploration delves into the nuanced ways in which randomness influences
various cognitive processes, shaping decision-making, learning, perception, and behavior.

Randomness in Cognitive Science:

1. **Stochastic Processes in Decision-Making:

Decision-making often involves navigating uncertain environments where outcomes are

probabilistic rather than deterministic. Stochastic processes, such as random walks and
Markov decision processes, capture the dynamic nature of decision-making under uncertainty,
allowing for the integration of randomness into cognitive models.

2. Probabilistic Learning Mechanisms:

Human learning mechanisms are inherently probabilistic, relying on random sampling and
statistical inference to update beliefs and acquire knowledge. Bayesian learning models, for
instance, leverage randomness to explore hypothesis spaces, estimate parameters, and adapt
to new information over time.

3. Noise and Variability in Perception:

Sensory perception is subject to inherent noise and variability, stemming from physiological
limitations, environmental factors, and neural processing mechanisms. Random fluctuations in
sensory signals influence perceptual judgments, response times, and the reliability of
perceptual inferences.
4. Random Exploration Strategies:

Cognitive systems often employ random exploration strategies to discover novel information,
solutions, or strategies in complex environments. Random exploration mechanisms enable
adaptive behavior, facilitating the discovery of optimal solutions and avoiding suboptimal
decision traps.

Methods for Exploring Random Computation:

1. Monte Carlo Simulation Techniques:

Monte Carlo simulation techniques harness the power of random sampling to estimate
complex probabilistic models, simulate stochastic processes, and evaluate the robustness of
cognitive models. These simulations provide insights into the range of possible outcomes and
the uncertainty inherent in cognitive tasks.

2. Randomized Experimental Designs:

Randomized experimental designs are employed to introduce controlled randomness into

cognitive experiments, ensuring that observed effects are not confounded by systematic biases
or external factors. Random assignment of participants to experimental conditions helps isolate
causal relationships and generalize findings.

3. Stochastic Model Fitting and Inference:

Stochastic modeling approaches involve fitting probabilistic models to empirical data and
performing inference to estimate model parameters, assess model fit, and make predictions
about cognitive processes. Techniques such as maximum likelihood estimation and Bayesian
inference leverage randomness to explore model spaces and quantify uncertainty.

4. Agent-Based Modeling and Simulation:

Agent-based modeling and simulation frameworks simulate the behavior of autonomous

agents in complex environments, incorporating random elements to capture individual
variability, interaction dynamics, and emergent phenomena. These simulations allow
researchers to explore the collective behavior of cognitive systems and study the effects of
randomness on social phenomena.
Applications in Cognitive Science:

1. Robust Decision-Making Under Uncertainty:

Understanding how individuals make decisions in uncertain contexts is crucial for predicting
behavior, designing interventions, and optimizing decision support systems. Random
computation techniques help elucidate the cognitive processes underlying robust decision-
making strategies, risk perception, and adaptive behavior in uncertain environments.

2. Adaptive Learning Strategies:

Exploring random computation sheds light on the mechanisms underlying adaptive learning
strategies, such as reinforcement learning, exploration-exploitation tradeoffs, and curiosity-
driven exploration. These insights inform the development of computational models that
capture the flexible nature of human learning and decision-making.

3. Statistical Analysis of Behavioral Data:

Random computation methods play a pivotal role in the statistical analysis of behavioral data
collected in cognitive experiments. Techniques such as bootstrapping, permutation tests, and
Monte Carlo simulations allow researchers to account for variability, assess the significance of
observed effects, and derive robust conclusions from empirical data.

Conclusion:

Exploring random computation in cognitive science unveils the intricate relationship between
randomness and human cognition, providing a rich tapestry of insights into decision-making,
learning, perception, and behavior. By embracing randomness as a core feature of cognitive
processes, researchers can unravel the complexities of the mind and pave the way for
innovative approaches to understanding and enhancing human cognitive performance.

5) Coroutines: Functions that Receive Continuations in Cognitive Science

Introduction:

Coroutines, also known as cooperative multitasking, present a unique programming paradigm

where functions have the capability to suspend execution and transfer control to other
functions, known as continuations. In cognitive science, coroutines offer a versatile abstraction
for modeling cognitive processes that entail asynchronous or interleaved execution, such as
attentional shifts, task switching, and decision-making under uncertainty.
Coroutines in Cognitive Science:

1. Modeling Attentional Mechanisms:

Coroutines serve as a valuable tool for simulating attentional mechanisms within cognitive
tasks. By representing attention as a coroutine, researchers can dynamically simulate
attentional shifts and resource allocation. This flexibility allows for the modeling of adaptive
behavior in cognitive architectures, reflecting the dynamic nature of attentional processes in
human cognition.

2. Task Switching and Cognitive Flexibility:

Cognitive flexibility, characterized by the ability to switch between different tasks or mental
states, can be effectively modeled using coroutines. By defining tasks as coroutines with
associated continuations, researchers can simulate task switching behavior and investigate the
cognitive processes underlying cognitive flexibility. This approach enables the exploration of
factors influencing task-switching efficiency and adaptability.

3. Decision-Making under Uncertainty:

Coroutines provide a natural framework for modeling decision-making processes that involve
uncertainty and partial information. By suspending execution and invoking continuations based
on available evidence, cognitive models can simulate sequential decision-making in complex
environments. This capability allows researchers to explore the mechanisms underlying
adaptive decision-making strategies in uncertain conditions.

4. Temporal Dynamics of Cognition:

Coroutines facilitate the modeling of temporal dynamics in cognitive processes, including

learning, memory retrieval, and perceptual processing. By interleaving multiple coroutines
representing different cognitive processes, researchers can capture the dynamic interaction
between mental states over time. This approach enables the simulation of real-time cognitive
processes and the investigation of temporal dependencies in cognitive tasks.

Implementation and Applications:

1. Programming Paradigm:

Coroutines can be implemented in various programming languages using language features

such as generators, async/await syntax, and delimited continuations. These language
constructs provide the foundation for expressing coroutines and continuations within cognitive
modeling frameworks.
2. Cognitive Modeling Frameworks:

Cognitive scientists can leverage existing cognitive modeling frameworks to implement

coroutines and continuations in cognitive models. Frameworks such as ACT-R, Soar, and
PyACT-R offer tools for representing cognitive processes and integrating coroutines into larger
cognitive architectures. This integration enables the development of comprehensive models
that capture the dynamic nature of cognition.

3. Experimental Design and Simulation:

Coroutines enable researchers to design experiments and simulations that reflect the
dynamic nature of cognitive processes. By embedding coroutines within experimental
paradigms, researchers can test hypotheses about attention, task switching, and decision-
making in controlled environments. This approach facilitates the investigation of cognitive
processes in real-time and the validation of cognitive models against empirical data.

Challenges and Considerations:

1. Complexity and Cognitive Load:

Implementing coroutines in cognitive models may introduce additional complexity and

cognitive load. Careful design and validation are necessary to ensure that coroutines accurately
capture cognitive processes without overwhelming cognitive resources. Researchers must
strike a balance between model complexity and cognitive plausibility to maintain model
interpretability.

2. Interpretability and Validation:

Interpreting the behavior of coroutines and validating their predictions against empirical data
pose challenges in cognitive modeling. Researchers must develop methods for analyzing and
validating coroutine-based models using experimental techniques and behavioral measures.
This process involves comparing model predictions with observed behavior and iteratively
refining the model based on empirical evidence.

Conclusion:

Coroutines offer a promising framework for modeling cognitive processes that involve
asynchronous execution, attentional mechanisms, and decision-making under uncertainty. By
incorporating coroutines into cognitive architectures and experimental paradigms, researchers
can gain new insights into the temporal dynamics of cognition and develop more
comprehensive models of human behavior. Continued research in this area holds the potential
to advance our understanding of cognitive processes and inform the development of intelligent
systems and cognitive technologies.
6) Enumeration in Cognitive Science

Introduction:

Enumeration is a fundamental cognitive process that involves counting, listing, or

systematically examining items or possibilities. In cognitive science, understanding how
humans perform enumeration tasks provides insights into cognitive processes such as
perception, memory, problem-solving, and decision-making.

Cognitive Mechanisms of Enumeration:

1. Perceptual Enumeration:

- Humans can quickly and accurately enumerate small sets of objects without explicit
counting. This rapid enumeration ability is thought to rely on perceptual mechanisms that
extract numerosity information directly from visual scenes.

- Perceptual enumeration tasks involve subitizing, the ability to instantly recognize the number
of items in a small set (typically up to four or five items) without counting.

2. Serial Enumeration:

- Serial enumeration refers to the process of counting items one-by-one in a systematic

manner. It involves sequentially attending to each item in a set and incrementing a mental
counter.

- Serial enumeration is often used for larger sets of items that exceed the subitizing range. It
requires more cognitive effort and may involve strategies such as verbal counting or visual
scanning.

Applications of Enumeration in Cognitive Science:

1. Memory and Attention:

- Enumeration tasks are used to study the capacity and limitations of working memory and
attention. For example, researchers investigate how efficiently individuals can enumerate items
presented briefly in visual displays and how distractions or cognitive load affect enumeration
performance.

2. Numerical Cognition:

- Enumeration tasks contribute to our understanding of numerical cognition, including the

development of numerical skills in children and the impact of numerical abilities on
mathematical achievement.
 - Studies on enumeration errors and biases shed light on cognitive processes underlying
numerical estimation and judgment.

3. Decision-Making and Problem-Solving:

- Enumeration is relevant to decision-making and problem-solving tasks that involve

evaluating multiple options or generating alternative solutions.

- Understanding how individuals enumerate possibilities and weigh alternatives informs

theories of decision-making under uncertainty and rationality.

Methodological Considerations:

- Researchers use various experimental paradigms to study enumeration, including visual

search tasks, enumeration of arrays, counting tasks, and response time measures.

- Computational models of enumeration incorporate cognitive processes such as visual

processing, attentional allocation, and working memory capacity to simulate human
performance in enumeration tasks.

Challenges and Future Directions:

- Future research in enumeration aims to elucidate the neural mechanisms underlying

enumeration processes, integrate computational models with neuroimaging data, and develop
interventions to improve numerical skills and decision-making abilities.

- Challenges include investigating individual differences in enumeration performance, exploring

cultural and developmental influences on enumeration strategies, and extending research
findings to real-world applications such as education and healthcare.

Conclusion:

Enumeration is a multifaceted cognitive process that plays a crucial role in various domains of
cognitive science. By studying how humans enumerate items and possibilities, researchers gain
insights into the mechanisms underlying perception, memory, decision-making, and numerical
cognition, advancing our understanding of human cognition and behavior.