The secret of the parallel distributed processing model: How does the brain simulate neural networks?

In psychology, parallel processing is the brain's ability to process stimuli of varying quality simultaneously. This ability is specifically related to the visual system, as the brain separates what we see into four components: color, motion, shape, and depth, analyzes each separately, and then compares them with stored memories, which helps the brain recognize what we are looking at. All of this information is seamlessly combined into the vision we see and understand.

For example, when a person stands in a crowd and two groups of people are having different conversations at the same time, he may only be able to hear some information from both conversations at the same time.

The concept of parallel processing has been linked by some experimental psychologists to the Stroop effect. The Stroop effect is derived from the Stroop test, which demonstrates the mismatch between color names and actual colors. In the Stroop effect, people's selective attention prevents them from processing all stimuli at once, leading to confusion.

Background

In 1990, American psychologist David Rumelhart proposed the parallel distributed processing (PDP) model, which aims to study neural processes through computer simulation. According to Rumelhart's theory, the PDP model views information processing as interactions between units, which can be either activating or inhibitory in nature. The parallel distributed processing model is inspired by the organizational structure of the nervous system and simulates the biological nervous system.

Parallel Processing vs. Serial Processing

In contrast to parallel processing, serial processing involves processing information in a sequential manner with no overlap between different processing times. The difference between these two processing models is most evident in the processing of visual stimuli.

Visual Search

In case of serial processing, elements are searched one by one in order to find the target. When the target is found, the search is terminated; if not, the search continues to the end to ensure that the target does not exist. In contrast, parallel processing processes all objects at the same time, although the completion time may vary.

However, the efficiency of parallel processing when facing some complex tasks is still a concern, which will be discussed in detail in the following articles.

Aspects of the Parallel Distributed Processing Model

There are eight main aspects of the parallel distributed processing model:

  • Processing units: include abstract elements such as features, shapes, and words, and are generally divided into three types: input units, output units, and hidden units.

  • Activation state: This represents the state of the system. The activation pattern is represented as a vector of real numbers.

  • Output function: maps the current activation state to the output signal.

  • Connection pattern: determines how the system responds to any input. The overall connection pattern is represented by the weight of each connection.

  • Propagation rule: Determine the net input of each type of input based on the output vector and the connection matrix.

  • Activation rule: The activation state of all units produces a new state based on the net input and the current activation state.

  • Learning rules: Modify connection patterns through experience, including establishing new connections, losing existing connections, or modifying existing weights.

  • Environment representation: The environment is represented by a time-varying randomness function, that is, the input pattern that may be applied to the input unit at any point in time.

Depth perception

Humans perceive depth using their eyes. This perception is present at birth and is also possessed by some animals such as cats, dogs and monkeys. Certain visual cues help us establish depth perception, including binocular cues and monocular cues.

Limitations of Parallel Processing

Limitations of parallel processing were raised in several analytical studies. The main limitations include brain capacity constraints, interference from attentional flicker lapses, limited processing power, and information limitations in visual search. For complex tasks such as object recognition, it is impossible for the brain to fully process them all in parallel.

Feature Integration Theory

Anne Treisman's feature integration theory integrates serial and parallel processing and takes into account the allocation of attentional resources. The theory is divided into two stages: feature detection using parallel processing and integration of features using serial processing, helping us form the perception of holistic objects and patterns.

The parallel distributed processing model reveals many secrets about the operation of neural networks, but the complexity and limitations of this process still make us think: How does the brain balance these processing modes to achieve our daily perception and response?

Trending Knowledge

The brain's superpowers: How does parallel processing analyze different stimuli simultaneously?
Parallel processing is an important ability of the human brain that allows stimuli of varying quality to be processed simultaneously. This ability is particularly evident in the visual system, where t
nan
With the advancement of medical technology, peritoneal dialysis (PD) has gradually become an important choice for care for patients with renal failure.According to the latest research, compared with t
Mysteries of the visual system: How do our eyes resolve color, motion, shape, and depth?
In our daily lives, our eyes seem to capture and parse the world around us in a seamless way. However, this seemingly simple process involves the brain's complex parallel processing system. According

Responses