PSIHOLOGIE COGNITIVĂ

Introducere Bazele psihologiei cognitive

Resurse necesare și recomandări de studiu 11

Durată medie de parcurgere a unității 11

La sfârşitul parcurgerii acestei unităţi, studenţii vor fi capabili să:

- Le fie familiare problematica şi specificul domeniului psihologiei cognitive.

- Sublinieze momentele importante din istoria scurtă a apariţiei psihologiei cognitive.

- Evidenţieze criticile şi meritele acestui domeniu

Psihologie generală (mecanismele cognitive), introducere ợn filosofie/logică, informatică.

Resurse necesare și recomandări de studiu

1 Miclea, M., (1997 sau 2000), Psihologie cognitivă, Polirom, Iaşi

2 Zlate, M., (1999), Psihologia mecanismelor cognitive, Polirom, Iaşi

Durata medie de parcurgere a unității de studiu

Cognitive psychology has evolved into an applied field closely linked to cognitive sciences, information technology, artificial intelligence, cybernetics, logic, language theory, and neuroscience These interconnected disciplines extend into practical areas such as engineering, construction, and design As a result, cognitive psychology encompasses a broad and diverse research landscape, developing its own unique language distinct from traditional psychological disciplines The technical nature of this field can initially surprise students enrolled in the course, often leading to the term "cybernetic psychology" or "psychology as a computer." The striking similarities between information technology and the way this branch of psychology describes the human psyche can be challenging for students accustomed to viewing the psyche as inherently "human" rather than mechanical.

Cognitive psychology explores the human cognitive system, focusing on themes such as perception, thought, memory, and language, along with its subsystems This cognitive system is viewed as an information processing entity that transforms sensory input into motor or behavioral output Similar to a computer, the mind has inputs and outputs, responding to stimuli, which aligns with principles found in behaviorism.

Cognitive psychology distinguishes itself from other branches of psychology by examining psychological phenomena through the lens of informational mechanisms that govern their functioning It aims to depict the human psyche as a system capable of exchanging information with its environment, transforming data based on its own rules and processing subsystems, and exhibiting self-regulation and control, closely linking it to cybernetics The origins of cognitive psychology can be traced back to the mid-20th century, following World War II, when mathematical logic, cybernetics, and information theory gained prominence in the international scientific landscape This effort to formalize logic enabled the manipulation of symbols based on general and recursive syntactic rules, leading to the definitions of computability and computation.

Alături de teora informaţiei şi logica matematică şi simbolică, curentele psihologice ale căror idei şi cercetări psihologia cognitivă le-a integrat sunt asociaţionismul, gestaltismul, behaviorismul şi constructivismul piagetian

Instrumentarul metodologic folosit ợn cercetare conţine metoda experimentului ca metodă predilectă, la care se adaugă modelarea-formalizarea- simularea pe calculator.

The field of cognitive science began to take shape in the 1950s, but its theoretical roots date back to ancient Greece with Aristotle, the founder of logic, and later to René Descartes, who famously questioned reality and concluded that the only certainty is our ability to think, as expressed in his phrase "cogito." The development of cognitive science gained momentum during World War II, when mathematician and logician Alan Turing built the theoretical Turing machine, a universally computable device that demonstrated a machine's ability to solve any mathematical calculation represented algorithmically, laying the groundwork for modern computer science.

Alan Turing is often regarded as the "father" of modern computers, including tablets He proposed a method for evaluating artificial intelligence, suggesting that if a system's responses are indistinguishable from those of a human observer, it can be deemed intelligent By 1956, two key research groups focused on information processing mechanisms had emerged: one at MIT and the other at Carnegie-Mellon That fall, MIT hosted a symposium, and on September 12, recognized as the birth date of cognitive sciences, three landmark presentations were made: H.A Simon and A Newell showcased the first computer demonstration of a logical theorem (Logic Theory Machine), while N Chomsky introduced his work on Three Models of Language.

In his foundational work on theoretical linguistics, G Miller introduces his renowned study "The Magical Number Seven, Plus or Minus Two." However, it is Ulric Neisser who popularizes the term "cognitive psychology" with the publication of his book of the same name in 1967.

The Alfred P Sloan Foundation significantly boosted the advancement of cognitive sciences by granting hundreds of millions of dollars in research funding during the mid-1970s, prompting many researchers to radically alter their programs to access these funds In 1978, the first research report, known as the SOAP (State of the Art Paper), was presented to the Sloan Foundation, stating, "the existence of our research domain is determined by a common goal: discovering the computational and representational capabilities of the mind and their structural and functional projections in the brain." This "declaration of independence" emphasizes that cognitive sciences study both natural and artificial cognitive systems, seeking to explain how information is processed in these systems in terms of computation and representation, as well as the implementation of these operations in various physical environments available at the time.

Cognitive psychology is an integral part of the cognitive sciences, which also include philosophy (epistemology, philosophical logic, analytic philosophy), linguistics (theoretical linguistics), neuroscience (cognitive neuroscience), and anthropology (cultural anthropology) These fields contribute to cognitive sciences not in their entirety but through specific branches and interdisciplinary developments.

Cognitive psychology faces dual pressures that give it a bifocal character Top-down pressures from artificial intelligence require cognitive psychology to provide formalized and computable models Conversely, bottom-up pressures from neuroscience demand that cognitive psychology develop valid, relevant, and predictive models of human behavior.

Noul domeniu are ợnsă şi critici Trei categorii de critici s-au conturat de-a lungul timpului:

Critics argue that there is a lack of neuronal plausibility in cognitive models, questioning the ecological validity of these theories They find it challenging to identify the neuronal counterparts of various cognitive processes and rules described in cognitive modeling For instance, they ask where the "if then" production rules are located within the cognitive system The cognitive system is analyzed at four levels: semantic, algorithmic-representational, implementational, and computational However, these critics often confuse these levels by demanding an implementational analysis of entities that are fundamentally at the computational level Furthermore, most cognitive models incorporate data from neuroscience, particularly within the neoconnectionist paradigm.

Behaviorists have criticized the methodologies employed by cognitive psychologists, particularly the reliance on computer modeling and simulation to achieve research results They question the ability of cognitive factors to influence behavior However, these criticisms have diminished in recent years, as many experimental findings from behaviorism have been integrated into cognitive psychology Additionally, behaviorists are now represented in the theoretical frameworks of connectionist models of cognitive systems.

Some critics argue that cognitive psychology is merely a trend that will soon fade away However, we can counter this claim by stating that as long as the mind functions as an informational system, cognitive psychology will endure Given that the mind is fundamentally an information-processing system, it is likely to remain relevant for a long time.

Cognitive psychology provides a universal framework for understanding computational capabilities, information processing, and responses to environmental stimuli, extending beyond human subjects This theoretical approach opens avenues for comprehending internal cognitive structures and communicating with non-human subjects By formalizing information processing within cognitive sciences, we gain a broader perspective on life, emphasizing that all organisms require an external information processing system and adaptive response mechanisms for survival Cognitive sciences encourage us to understand other subjects and think beyond our own humanity.

Cognitive psychology is a rapidly expanding and developing interdisciplinary field that primarily relies on experimental research It lies at the intersection of simulating and implementing human cognitive functions through computer models and examining the neuropsychological and behavioral plausibility of supported theories Since its inception in the 1950s, marked by key publications such as H.A Simon and A Newell's demonstration of a logical theorem on a computer and N Chomsky's "Three Models of," cognitive psychology has continuously faced various critiques and challenges.

Language, G Miller - The Magical number seven, plus or minus two.) Cel care consacră termenul de „psihologie cognitivă” este Ulric Neisser, care publică ợn 1967 o carte cu acelaşi titlu.

Cuvinte cheie Ştiinţe cognitive Psihologie cognitivă

1 Care este metoda predilectă de cercetare a psihicului ợn cadrul psihologiei cognitive? (pg 12)

(D) metoda investigării pe bază de chestionar.

2 Obiectivul de cercetare al psihologiei cognitive este (p 11):

(A) Studierea capacităţilor de calcul, reprezentarea şi acţiune ale psihicului.

(B) Studierea capacităţilor analitice şi reprezentaţionale ale psihicului precum şi studirea creierului, structural şi funcţional.

(C) Studirea capacităţilor de reprezentare şi de proiectare ale psihicului, funcţionarea şi capacitatea creierului.

(D) Studierea capacităţilor computaţionale şi de reprezentare ale psihicului precum şi ale proiecţiilor lor structurale şi funcţionale ợn creier.

3 Care sunt criticile aduse de behaviorişti psihologiei cognitive? (p 14)

4 Ce curente psihologice au contribuit la apariţia psihologiei cognitive? (p 12)

5 Prezentaţi ce se ợnţelege prin „dubla presiune” (p.13).

Domeniul psihologiei cognitive este robust, puternic, fundamentat pe cercetări experimentale, fiind unul dintre domeniile cu o promiţătoare dezvoltare viitoare.

Nivelurile de analiză ale sistemului cognitiv

Nivelurile de analiză ale sistemului cognitiv 20

La sfârşitul acestei prelegeri, studentul va putea :

- Aibă o altă pespectivă asupra sistemului cognitiv;

- Să ợnţeleagă că un sistem cognitiv poate fi analizat din punctul de vedere a celor patru nivele;

- Să utilizeze noile informaţii pentru a ợnţelege perspectiva cognitivistă asupra psihicului.

Resurse necesare și recomandări de studiu.

Prelegerea care urmează va aborda una dintre temele importante ợn psihologia cognitivă, avõnd ca scop prezentarea sistemului cognitiv ợntr-o perspectivă nouă.

A cognitive system is defined as a physical system that possesses two key properties: representation and computation In other words, it is a physical entity that utilizes stored information in various forms, whether energetic or chemical To date, there is no evidence of a purely informational cognitive system that operates independently of a physical structure, with the brain serving as the supporting physical structure for the psyche.

Nu orice sistem fizic este şi cognitiv, ci doar acela care are capacităţi de reprezentare şi de calcul.

In cognitive psychology, representation is defined differently than in classical psychology, serving as an internal reflection of external reality rather than a mere sensory schematic of an absent object This concept encompasses a broader scope, referred to as mental imagery, which includes all information processed by the system and translated into its unique internal language This definition of representation, utilized in cognitive sciences, is derived from symbolic logic, as formalized by A Newell (1992).

Luăm un eveniment extern: variabila X se transformă (T) ợn variabila Y Rezultă formula X-T-Y (adică variabila X se transformă ợn variabilaY)

Reprezentarea evenimentului X-T-Y ợntr-un mediu intern se realizează cõnd o proiecţie x a variabilei X şi o proiecţie t a lui T ợn acest mediu pot genera o variabilă y care să corespundă lui Y

It is essential to establish a systematic relationship between external variables and their internal references This means that the representation of an object or event must consistently correspond to the same entity in the external environment, although it is important to note that these representations are not identical; they do not serve as direct copies of reality.

Reprezentările conţinute de un sistem cognitiv (orice sistem cognitiv) sunt simbolice

Imagery, signs, and symbolic content, as well as subsymbolic elements like neural network activation patterns, often lack semantic interpretation and meaning Don't stress about understanding these concepts; we will revisit them when discussing cognitive psychology paradigms.

Calculul este proprietatea sistemului cognitiv care constă ợn manipularea reprezentărilor pe baza unor reguli

If representations are symbolic, we adhere to rules for manipulating symbols, including mathematical operations and grammatical, syntactic, and semantic rules Conversely, if representations are subsymbolic, we follow rules for modifying activation states, such as Hebb's rule and the generalized delta rule, which will be discussed further.

Corespunzător celor două tipuri de reprezentări şi reguli de manipulare ale lor vom avea două tipuri de modelări (teorii) ợn ştiinţele cognitive: clasic-simbolice şi neuromimetice

Orice sistem cognitiv este organizat ợn patru niveluri care care ợi definesc structura (Miclea, 2003, pp 28-35):

These are not layers or tiers of a cognitive system; instead, they are integrated and operate at the same execution level In other words, information is not transferred from one level to another for processing, as the four levels function simultaneously in time and are non-hierarchical in terms of importance.

Iată la ce se referă cele patru niveluri de analiză:

La acest nivel de descriere comportamentul sistemului cognitiv este ợnţeles pe baza volumului şi naturii cunoştinţelor pe care le are, a scopurilor şi intenţiilor sale.

A person's behavior, shaped by their cognitive system, is driven by their goals and the knowledge they possess For instance, individuals interpret phenomena like meteor showers differently based on their understanding; one may attribute it to divine intervention, while another might explain it through scientific reasoning related to gravitational forces Knowledge sources include immediate or long-term experiences, problem statements, skills, social interactions, and cultural participation.

Knowledge can be categorized into cognitive-penetrable and cognitive-impenetrable types Cognitive-penetrable behaviors and mental mechanisms are those that change based on an individual's existing knowledge (Pylyshyn, 1984, 1990).

Recognizing a letter is easier when it appears in a word rather than in a random combination of letters Similarly, if I were informed of a flood occurring on the floor above the classroom, I would likely choose to leave immediately.

Cognitive impenetrability refers to behaviors that are not influenced by an individual's knowledge For instance, the perception of an object's contours remains consistent regardless of what the person knows; they will process the contours of objects in the same manner in every situation, consistently recognizing the outlines of each perceived object.

It is important to distinguish between different types of information processing within the cognitive system This involves the flow of information from peripheral levels—such as color, texture, contours, movement, and sounds—toward central levels, which encompass semantic and functional aspects, including the category to which an item belongs, its meaning, and its function for the subject This process is known as bottom-up analysis or data-driven processing.

Top-down analysis, also known as knowledge-driven processing, involves utilizing existing information within a system For instance, human facial features are more easily recognized when placed in the context of a human figure Similarly, when reading, we process information from the bottom up by recognizing the shapes and spatial arrangements of letters into words, and then we interpret the text from the top down, relying on our knowledge of the Romanian language to understand what we see.

The computational level of the cognitive system addresses various tasks and problems that require resolution One effective method for analyzing the cognitive system is task analysis, which involves breaking down the task into its components and specifying each cognitive mechanism involved in solving those components The primary goal at the computational level is to thoroughly establish the processes that the problem's data (input) undergoes to achieve the solution (output).

Computational analysis of the visual system reveals that a two-dimensional image of three-dimensional objects is projected onto the retina However, this projection could correspond to an infinite number of three-dimensional objects The system's task is to determine which three-dimensional object created that two-dimensional image At the computational level, it identifies the cognitive mechanisms that enable the subject to recognize the three-dimensional object (the output) that produced the pixelated two-dimensional image (the input) This processing relies on depth calculation using binocular disparity.

There are two types of processing: modular processing, which is not influenced by the subject's knowledge (e.g., primary processing of visual information), and non-modular processing, which is affected by what the subject knows (e.g., object recognition).

Analiza algoritmului care realizează funcţia input-output şi a modalităţilor ợn care sunt reprezentate input-ul şi outputul poartă numele de nivel algoritmic- reprezentaţional.

Algoritmul: o secvenţă de calcule pe baza căreia, printr-un număr finit de paşi din datele de intrare, se obţin datele de ieşire.

Reprezentarea: modul de codare a input-ului (semantic, imagistic, serial, prin valori de activare, etc.) Un anumit tip de reprezentare poate favoriza un anumit algoritm

Paradigmele psihologiei cognitive Paradigma clasic simbolică şi paradigma neoconexionistă

Paradigma clasic simbolică şi paradigma neoconexionistă

Paradigmele psihologiei cognitive Paradigma clasică simbolică şi paradigma neoconexionistă

- Să ợnţeleagă temele principale şi fundamentele psihologiei cognitive;

- Să aibă o perspectivă nouă asupra sistemului cognitiv uman;

- Să compare şi să ợnţeleagă diferenţele dintre cele două paradigme.

2 Zlate, M., (1999), Psihologia mecanismelor cognitive, Polirom, Iaşi.

Cognitive psychology currently features two main paradigms for understanding and analyzing cognitive systems: the symbolic paradigm, also known as the classical or computational model, and the (neo)connectionist paradigm.

In the study of cognitive systems, two main paradigms are identified based on the type of representations and their processing methods: the classical-symbolic paradigm and the neoconnectionist paradigm These frameworks provide distinct approaches to understanding cognitive functions and the underlying mechanisms of representation.

The classical symbolic paradigm has significantly influenced the development of cognitive psychology and cognitive sciences as a whole, being the first to emerge chronologically It is based on the concept that all knowledge and states of affairs are represented in the cognitive system through symbols or symbolic structures.

Symbols are representations that denote objects or states and follow specific combinatory rules, akin to a "grammar" that transcends the knowledge or statements they symbolize For instance, the expression p→q indicates a causal relationship: if p (the light bulb is turned on), then q (there is light in the room) This system encapsulates real-world causality in its own language, where p represents the light bulb and q signifies the light The implication, a relationship that lacks meaning without its terms, is illustrated by an arrow Notably, the terms can be replaced with an infinite array of objects while maintaining the relationship, making it applicable to countless situations From a young age, children can extrapolate this relationship as soon as they grasp it.

The theory provides insights into the functioning of the human mind as both a symbolic and functional structure It suggests that high-level psychological processes such as abstraction and reasoning can be understood as the manipulation of symbols.

To effectively process knowledge, both computers and the human brain must encode information into symbolic structures Newell and Simon assert that both the human cognitive system and computers are physical-symbolic systems They are considered physical because they are instantiated neurobiologically, and symbolic because they represent information in the form of symbols or symbolic expressions that aggregate according to specific rules or functions.

The emergence of these theories was bolstered by the successes in formalizing logic, the development of generative grammars, and the advancements in theoretical linguistics, which collectively led to the notion that the human psychological system functions as a mechanism for manipulating symbols.

The classical symbolic paradigm has been particularly effective in explaining central cognitive processes, especially in problem-solving, which involves the initial state, final state, and the operator block that facilitates the transition from the initial to the final state.

Researchers in this paradigm aim to develop computer programs that mimic or simulate aspects of human cognitive functioning to better understand human cognition, a process known as computational modeling Unlike cognitive science, artificial intelligence focuses on building systems that generate intelligent responses, although the information processing methods may differ significantly from human processes, as defined by A Turing's concept of intelligent systems This model allows for the discussion of the computer metaphor in explaining the human cognitive system.

Ea se numeşte “clasică” deoarece foloseşte idei dezvoltate ợn lucrările unor filosofi cum ar fi Leibniz, Descartes (raţionalismul continental) sau Hobbes, Locke, Hume (empirismul englez)

Arhitecturile cognitive (AC) ẻn cadrul realizării unei arhitecturi cognitive accentul cade pe intracţiunea dintre structuri.

Termenul de AC este ợmprumutat din inteligenţa artificială.

Artificial intelligence (AI) refers to the branch of information technology focused on creating intelligent cognitive structures that do not necessarily resemble human ones Computer modeling involves the implementation of human-specific cognitive functions and structures on a computer to enhance understanding of these processes.

AC sunt totalitatea mecanismelor, cognitiv impenetrabile (vezi Pylyshyn, 1984,

1990) necesare şi suficiente pentru realizarea unui comportament inteligent (deci sunt „comuter modelling”)

Sunt aspecte ale sistemului cogitiv care rămõn ợn cea mai mare parte invariabile de la un subiect la altul, la tipuri de sarcini şi ợn timp.

La baza lor stau sistemele de producere şi regulile de producere (production systems and production rules): condiţionali şi sisteme de condiţionali de tipul „dacă atunci”.

Newell şi Simon (1972) au fost primii care au relevat valoarea sistemelor de producere ợn rezolvarea de probleme.

ACT-R (Adaptive Control of Thought - Rational) is a cognitive architecture developed by J.R Anderson between 1993 and 2008 It serves as a theoretical framework for simulating and understanding human cognition, focusing on how individuals organize knowledge and exhibit intelligent behavior The ongoing research aims to refine ACT-R into a comprehensive system capable of performing a wide array of human cognitive tasks, detailing the processes of perception, thought, and action For more information, visit the official website: [ACT-R](http://act-r.psy.cmu.edu/).

Este de inspiraţie computaţională şi dezvoltă un model general al funcţionării cognitive umane şi a rezolvării de probleme, ợn special.

Asumpţiile (presupoziţiile) majore ale teoriei ACT-R sunt:

1 Sistemul cognitiv are şase module care funcţionează relativ independent unul de celelalte.

2 Patru sunt importante pentru cogniţie şi pentru rezolvarea de probleme:

- modulul de recuperare: conţine indiciile sau cheile de accesare a informaţiilor Este localizat ợn cortexul prefrontal inferior ventrolateral.

- modulul imagistic: transformă problema ợn reprezentare vizuală Este situat ợn cortexul parietal posterior.

- Modulul scopului: urmăreşte intenţiile unui subiect şi controlează procesarea informaţiilor Este localizat ợn cortexul cingulat anterior

- Modulul procedural: utilizează regulile de producere, daca atunci, pentru a determina care va fi următoarea acţiune ợntreprinsă Se află ợn ganglionii bazali, ợn capătul nucleului caudal

- Celelalte module sunt: modulul motor şi modulul vizual.

3 ariile cerebrale corespunzătoare celor patru module sunt activate ợn general de sarcini complexe, cu toate că fiecare este activată de factori realativi diferiţi.

4 fiecare modul are un buffer asociat lui, care conţine o cantitate limitată de informaţie

5 cunoaşterea umană poate fi redusă la două tipuri de reprezentări: declarative şi procedurale.

Această arhitectură, implementată pe calculator, poate rula pe oricare calculator, adică poate rezolva probleme, dacă se descarcă codul-sursă şi calculatorul deţine limbajul Common Lisp.

- 4CAPS, dezvoltat la Carnegie Mellon University sub coordonarea lui Marcel A Just

- ACT-R, dezvoltat de Carnegie Mellon University sub coordonarea lui John R

- Apex dezvoltat de Michael Freed la NASA Ames Research Center.

- CHREST, dezvoltat de Fernand Gobet la Brunel University şi de Peter C Lane la University of Hertfordshire.

- CLARION the cognitive architecture, dezvoltat de Ron Sun la Rensselaer Polytechnic

Institute şi la University of Missouri.

- Copycat, al lui Douglas Hofstadter şi Melanie Mitchell la Indiana University.

- DUAL, dezvoltat de New Bulgarian University sub coordonarea lui Boicho Kokinov.

- EPIC, dezvoltat de David E Kieras and David E Meyer la University of Michigan.

- FORR dezvoltat de Susan L Epstein la The City University of New York.

- GAIuS dezvoltat de Sevak Avakians.

- CoJACK o arhitectură inspirată de ACT-R, JACK, un multi-sistem care adaugă o arhitectură cognitivă pentru a simula ợn mediul virtual comportamente asemănătoare cu cele umane.

- PreAct, dezvoltat de Dr Norm Geddes la ASI.

- PRODIGY, al lui Veloso et al.

- PRS 'Procedural Reasoning System', dezvoltat de Michael Georgef şi Amy

- Psi-Theory dezvoltat de Dietrich Dửrner la Otto-Friedrich

- R-CAST, dezvoltat de Pennsylvania State University.

- Soar, dezvoltat de Allen Newell and John Laird la Carnegie Mellon University şi la the University of Michigan.

- Society of mind and şi succesoarea sa Emotion machine dezvoltate de Marvin

- Subsumption architectures, dezvoltate de Rodney Brooks (ợncă se dezbate dacă sunt totuşi cognitive).

Este cunoscută şi sub numele de paradigma procesărilor paralele distribuite sau de modelarea neuromimetică.

Porneşte de la ideea că activitatea cognitivă poate fi explicată pe baza unor modele de inspiraţie neuronală Un sistem cognitiv este format dintr-o reţea de neuroni formali (simplificaţi).

The initial attempts to implement the concept of neural networks were made in 1943 by Pitts and McCullough Building on their research, Rosenblatt developed a two-layer neural network known as the perceptron in 1958 However, Minsky and Papert's 1969 studies revealed the limitations of perceptrons in computing simple logical functions such as "and," "or," and "exclusive or," leading to a decline in further development in this area In the 1980s, McClelland and Rumelhart revitalized the field with their publications, "Parallel Distributed Processing: Exploration in the Microstructure of Cognition" (1986) and "Psychological and Biological Models" (1987), marking the emergence of neoconnectionism as we know it today.

Unlike the symbolic paradigm, which views knowledge as the manipulation of symbols, connectionism asserts that information representation in the cognitive system involves the activation of simple units (neuron-like structures) based on patterns and activation values.

The cognitive system functions as a neuromimetic network that emulates the structure of neural networks, where information flows between processing units not as messages, but as activation values This process involves scalar representations rather than symbolic ones (D.A Norman, 1986).

Neoconexionismul susţine o modelare a procesării la nivelul algoritmic- reprezentaţional (a se vedea cursul despre nivelurile de analiză ale sistemului cognitiv) printr-o reţea neuromimetică.

Informaţia este reprezentată la nivelul sistemului cognitiv uman prin valori şi patternuri de activare ale unor unităţi (neuromimi).

O reţea neuromimetică este formată din:

5 un pattern de conexiuni ợntre aceste unităţi;

7 un mediu (o ambianţă) ợn care operează reţeaua

Să le luăm pe rând pentru a le explica.

Cognitive units, also known as formal neurons or nodes, exhibit characteristics similar to biological neurons, including an activation value that ranges from -1 to +1 In a network, there are input units that receive and convert external signals into activation values, and output units that convey the network's response to the environment Input and output units are visible and can be accessed directly, while hidden units, which cannot be accessed directly, are positioned between these visible units.

Procesarea informaţiei vizuale

- Să aibă o nouă perspectivă asupra analizatorului vizual;

- Să ợnţeleagă modelarea pe calculator a proceselor cognitive;

- Să fie familarizaţi cu teroriile recunoaşterii obiectelor.

Nearly 50% of the neocortex is dedicated to processing visual information, highlighting the significance of visual data in our understanding of the external environment This emphasizes the crucial role that visual information processing plays in any analysis of cognitive systems.

Modele dezvoltate ợn psihologia cognitivă pentru a explica acest tip de procesare trebuie să aibă atõt plauzibilitate neuronală, cõt şi eficienţă ợn implementarea lor ợn reţele non-umane

Numim un fenomen ca fiind vizibil atunci când emite unde electromagnetice cu lungimi de undă ợntre 440 şi 810 milimicroni şi care provoacă o activitate fotochimică la nivelul receptorilor.

Light penetrates the transparent layers of the eye and reaches the retina, which consists of five cellular layers: photoreceptors (cones and rods), horizontal cells, bipolar cells, amacrine cells, and ganglion cells The ganglion cells extend into axons that form the optic nerves.

The retina has a lattice-like structure where horizontal connections exist between nerve cells, facilitated by amacrine and horizontal cells Additionally, vertical connections are formed with bipolar cells, ganglion cells, and optic nerves, allowing for effective signal transmission.

Fig 1 Structura retinei (sursa www.wikipedia.com)

Această structură arată că nu se vizează procesarea ợntregii informaţii vizuale, ci doar a celei semnificative Se face o diferenţere şi o procesare segregată ợncă de la nivelul periferic al sistemului vizual.

Pe ansamblu, unei celule ganglionare ợi revin 120-130 de receptori

Informaţia este trimisă către nucleii geniculaţi laterali din talamus, apoi către cortexul vizual (striat) unde este analizată şi se produce recunoaşterea obiectului.

Câmp receptor, celulele on-off şi off-on

The area on the retina that alters the activity of nerve cells is known as the receptive field This occurs because multiple receptors correspond to a single ganglion, which does not activate in response to stimulation at just one point on the retina, but rather to stimulation across a broader area.

Este o noţiune funcţională, nu una biologică, adică nu există o formaţiune neuronală care să reprezinte câmpul receptor, aşa cum este retina, spre exemplu

On-off cells enhance the rate of electrical discharges, increasing action potential frequency when a light stimulus is presented at the center of their receptive field, while decreasing neuronal activity when the stimulus is located at the periphery Conversely, off-on cells exhibit peak activity when a dark spot is positioned at the center of their visual field, surrounded by beams of light.

The activity of two types of cells is unaffected by uniform stimulation of the receptive field due to lateral inhibition The presence of these cell types, located in the lateral geniculate bodies, can be demonstrated through neurophysiological tests and psychophysical methods, such as the Hermann-Hering grid.

Fig 2 Grilajul lui Hermann (1870) Privirea grilajului produce iluzia unor pete cenuşii la intersecţia canalelor albe situate ợntre careurile negre

Hubel and Wiesel, renowned researchers in visual processing, identified three types of feature detectors: simple cells, complex cells, and hypercomplex cells (Hubel & Wiesel, 1959).

Simple cells are specialized neurons that detect contours, bright edges, and lines, responding exclusively to one type of stimulus Each simple cell integrates the activity of multiple on-off and off-on cells from the geniculate nuclei.

Celulele simple sunt organizate somatotropic, adică o anumită regiune din cortexul vizual procesează informaţia dintr-o anumită zonă a câmpului vizual, ignorând stimulii dintr-o altă locaţie.

Complex cells are located in the striate and parastriate areas of the brain and process information with a higher level of generality They respond to the same category of stimuli regardless of their position in the visual field and are capable of decoding moving stimuli.

Hypercomplex cells are categorized into two types The first type responds to stimuli similar to those processed by complex cells but is sensitive to specific dimensions The second type is highly selective, detecting angles; some cells respond exclusively to right angles, while others are attuned to acute angles.

Unele cercetări sugerează că ar exista celule hipercomplexe care detectează chiar şi forme de genul triunghiului, mâinii sau chiar feţei

It is unlikely that complex shapes, such as faces, can be detected by a single cell If this cell were to die, the ability to recognize certain objects, like hands and familiar faces, would also vanish.

Hubel and Wiesel propose a hierarchical organization of three types of cells, yet they provide limited information about this arrangement It is possible, however, that all three cell types may operate simultaneously without transmitting information to one another in a stepwise manner.

Ne vom afla la nivelul computaţional, adică vom ợncerca să explicăm prin modele logico-matematice cum un input specific este calculat pentru a oferi un output specific.

Cu alte cuvinte, există o multitudine de algoritmi care să calculeze o funcţie şi o multitudine de sisteme fizice (de harduri) capabile să implementeze un algoritm.

Astfel, teoriile computaţionale care explică analiza vizuală a informaţiei sunt abstracte şi formalizate, ceea ce pare nenatural pentru psihologii tradiţionalişti.

La nivel computaţional se descriu două niveluri ale procesării: procesarea primară şi procesarea secundară a informaţiei

Procesarea primară cuprinde prelucrări pre-atenţionale cu o durată de aproximativ

200 milisecunde Au ca scop reprezentarea ợn sistemul cognitiv a caracteristicilor fizice ale stimului

Aflăm unde este stimulul, dar nu şi ce este Avem mecanismele de detectare a contururilor, a texturii, a mişcării, a culorii şi a dispunerii spaţiale.

Procesarea secundară vizează recunoaşterea obiectelor Are ca input rezultatele procesării primare şi ca output imaginea tridimensională a unui obiect identificat.

David Marr (1982) oferă schema generală a procesării informaţiei vizuale Stimulii vizuali sunt transpuşi ợntr-o schiţă primară, care este supusă unui tratament modular prin care se stabileşte:

• dacă este ợn mişcare şi cum

According to David Marr's 1982 model, visual information processing occurs through modular and parallel mechanisms, resulting in an intermediate representation known as 2.5D This term reflects its intermediary nature between the two-dimensional image on the retina and the three-dimensional perception of the visual scene.

După procesarea 2 ẵ D şi pe baza inputului ei se realizează procesarea secundară, imaginea 3D şi recunoaşterea obiectului, procesarea secundară.

Procesării primare ợi corespunde decõt primele două niveluri (schema primară şi imaginea 2 ẵ D), iar imaginea 3D şi recunoaşterea aparţin procesării secundare

The modules are organized to operate in parallel, functioning as cognitive and non-inferential systems that cannot be broken down into sequences or steps, making it impossible to trace how information is processed.

2 Sunt independente de natura stimulului Vor procesa la fel indiferent ce procesează.

3 Au un caracter irepresibil, automat (deschidem ochii şi vedem, fie ca vrem fie că nu vrem; auzim sunetele fie că vrem, fie că nu vrem).

The article discusses three key assumptions about reality: the assumption of rigidity, which suggests we do not live in a pulsating universe; the assumption of uniqueness, indicating that a single point on the retina corresponds to one specific point in reality; and the assumption of continuity, which posits that a point near another analyzed point shares similar characteristics It remains unclear whether these assumptions are innate to our cognitive system or learned through experience.

Mecanismul atenţiei

- Să fie familiarizaţi cu teoriile ce explică mecanismul atenţiei;

- Să compare modele explicative ale mecanismului atenţiei şi să găsească punctele tari şi punctele slabe.

The significance of this psychological mechanism becomes evident when we consider that our sensory analyzers are bombarded with approximately 100,000 bits of information per second, while our conscious processing capacity is limited to only 25-100 bits per second.

Se impune prelucrarea selectivă a informaţiei determinată de caracteristicile mediului şi de caracteristicile sistemului cognitiv

Selectivitatea atenţiei are două sensuri:

1 selecţia itemilor sunt prelucraţi şi izolarea lor de ce care nu sunt procesaţi;

2 selectivitatea procesărilor pentru aceeaşi categorie de itemi (unii sunt prelucraţi mai profund decât alţii).

Prin urmare, atenţia este un mecanism de prelucrare selectivă a informaţiei

Principala formaţiune implicată este sistemul reticulat Acesta pregăteşte cortexul şi analizatorii pentru a răspunde la adecvat la stimuli.

Dacă este lezată formaţiunea reticulată, se produce coma, iar dacă este activată, atunci persoana este trezită (dacă doarme, spre exemplu)

The reticular activating system stimulates the cortex, which in turn can either excite or inhibit the activity of the reticular formation through descending pathways Located in the brainstem, the reticular substance generates tonic activity in the cortex, while its projection from the thalamus triggers a phasic response essential for switching and focusing attention.

Attention is a distributed psychological mechanism involving multiple neural components, including the anterior cingulate cortex for stimulus detection, the posterior cingulate cortex, and the reticular formation in the midbrain for stimulus tracking and attention shifting.

Research in neural pathology has demonstrated that primary processing is non-attentional, and the generation of 3D images remains unaffected by various lesions in areas responsible for attention mechanisms.

Neurochimia atenţiei a pus ợn evidenţă rolul pe care ợl au neurotransmiţătorii, catecolaminele pentru activare şi monoaminooxidaza pentru inhibare (MAO).

De reţinut: nu atenţia selectează informaţia care urmează să fie prelucrată mai profund, ci o serie de procesări ascendente (caracteristicile stimulului) şi descendente (valoarea adaptativă a stimului şi relevanţa sa motivaţională).

Modelarea atenţiei Modelul filtrajului timpuriu

The Broadbent model, proposed in 1958, suggests that information filtering occurs at the sensory level, likened to a "glass throat." According to this model, only a small portion of incoming information is processed, while irrelevant data is blocked by a filter that lacks any interpretative criteria.

Fig 8 Modelul filtrajului timpuriu, Broadbendt, 1958 (după Miclea, 2003)

MSD is not a location but rather a temporary state of activation of knowledge from MLD, as well as a conscious information processing system, as outlined in Baddeley and Hitch's working memory model.

The concept of attention switching is illustrated through dichotic listening experiments and "cocktail party" scenarios If we accept this model, it becomes challenging to explain how we can shift our focus to a message of interest that was not initially processed sensory, as noted by Broadbent.

Research has shown that filtering is not sensory but semantic, indicating that there is an interpretation of the meaning of information This was demonstrated in the dichotic listening experiments conducted by Gray and Wedderburn in 1960, where subjects wore headphones with a message starting in the right ear, stating "The sun was shining " and continuing in the left ear.

On a beautiful summer day, participants in an experiment were instructed to focus on a message delivered to either their right or left ear As the message concluded, it was replaced by nonsensical syllables until the second part of the message appeared After the experiment, subjects were able to reproduce the entire message However, if the filtering was sensory, they were expected to recall only the portion presented to the ear they were concentrating on.

Se trage concluzia că, cel puţin parţial, mesajele sunt prelucrate semantic la nivel inconştient

Este propus de Norman ợn 1968

Fig 9 Modelul filtrajului târziu (Norman, 1968, apud Miclea, 2003, p 116)

Attention functions like a workspace, akin to the surface of a table, where various neuropsychological activities compete Sensory stimuli are processed at a shallow level, activating long-term memory models, and the output of these processes is correlated with the relevance of the stimulus, determining its selection based on adaptive value This relationship between attention and stimulus relevance explains the shift of focus from less useful stimuli to those with greater adaptive significance Sensory processing occurs automatically, with environmental stimuli triggering feature detectors, while object recognition relies on temporarily activated long-term memory models, forming short-term (working) memory This activation is sustained through motivational relevance mechanisms that demand attentional allocation.

Mesajele procesate parţial sunt procesate la nivel subconştient tot semantic.

This model clarifies several phenomena that the previous model could not, such as the switching of attention to complete a dichotomous message However, it fails to explain why we cannot retain two different messages presented to the same ear at the same intensity and frequency, as one would expect no issues if the model were entirely accurate.

Anne Treisman's model, proposed in 1969 and 1988, hybridizes previous theories by suggesting that the filter does not operate on an all-or-nothing principle Instead, it attenuates signals, making successive selections at various levels of processing The processing of stimuli at any given moment is influenced by the current availability of the cognitive system.

Dacă sarcina din acel moment nu consumă pronunţat resursele sistemului, atunci din ce ợn ce mai mulţi stimuli din exterior pot intra sub procesare.

Dovezi de cercetare: intensitatea bioelectrică a procesării mesajului non- dominant din creier este mai mică decât intensitatea mesajului dominant.

Dezavantajul acestui model: disoluţia noţiunii de filtru Acesta devine multidimensional, adică ợntreg sistemul cognitiv acţionează ca un filtru.

Let's assume that every stimulus received by a subject is encoded by a neuromimetic network These stimuli create a neuromimetic network At a certain point, a stimulus is activated either due to its intensity or its motivational relevance, making it crucial for problem-solving.

Dacă postulăm că legătura dintre neuromimi se face pe baza unui mecanism similar inhibiţiei laterale, atunci când o unitate este activată, celelate sunt inhibate

Deci nici nu avem nevoie de postularea unor mecanisme de procesare atenţionale pentru a putea explica procesarea selectivă a unor stimuli, alegerea lor şi ignorarea altora.

The allocation of resources is crucial, as cognitive systems have limited processing capabilities Consequently, these systems must efficiently manage their restricted energy, employing lateral inhibition as a strategy to optimize performance.

Evidence of the interference phenomenon highlights performance deterioration when processing two tasks that require the same cognitive mechanisms The closer two units are in proximity, the more profound the lateral inhibition becomes For example, while we can see and hear from the television simultaneously, we cannot listen to two radios at the same time.

Sunt trei cauze ale interferenţei:

1 similaritatea stimulilor/sarcinilor de procesare/răspunsurilor solicitate;

2 nivelul de automatizare scăzut (sarcinile automatizate sunt mai puţin predispuse interferenţei);

3 nivelul ridicat de complexitate al sarcinii, cu cât sarcina este mai complexă, cu atât predispoziţia la interferenţă este mai mare

Limbajul

- Să fie familarizaţi cu teoriile recunoaşterii limbajului;

- Să ştie care sunt caracteristicile identificării cuvintelor;

- Să ştie care este deosebirea dintre procesarea primară şi procesarea secundară

3 Eysenck, M W., Keane, M.T (2010), Cognitive Psychology Sixth Edition, Psychology Press.

Literatura de specialitate şi cercetările desfăşurate ợn legătură cu acest subiect sunt străbătute de o idee centrală, aceea că recunoaşterea cuvintelor are un dublu input: buttom-up şi top-down

Aceasta ợnseamnă că recunoaşterea unui cuvõnt este determinată de datele exterioare care formează efectiv cuvântul, sunetele sau literele, cât şi de conceptele stocate care creează un context lingvistic mental.

Identificarea unui cuvânt are următoarele caracteristici:

1 Procesare automată (adică un cuvânt este procesat fără o prea mare implicare conştientă) A se vedea efectul Stroop.

2 Primaritatea semantică Dacă o propoziţie conţine cuvinte care adaugă un sens la un conext deja existent, atunci ele sunt mai repede recunoscute

3 Primaritatea cuvântului asupra literei Adică se identifică cuvântul, nu fiecare literă ợn parte, şi mai apoi cuvõntul (ex.: PSIHOLGIE)

Elementele de prozodie contribuie şi ele la identificarea cuvintelor (intonaţia, accentul pus pe anume cuvinte, topica, etc.).

Word recognition occurs even when the complete word or entire sentence is not received This is achieved through top-down processing, which fills in missing parts to derive meaning and significance.

The neurophysiological basis of language production begins with the auditory system The most crucial area of the brain for understanding speech is Wernicke's area, identified by Karl Wernicke in 1874 This region, located in the posterior and inferior portion of the left temporal lobe, plays a vital role in comprehending and formulating speech by transforming perceptual data into meaningful elements of words Damage to this area can result in Wernicke's aphasia, highlighting its significance in language processing.

The Broca area, identified by Paul Broca and located in the left frontal lobe, is crucial for the articulation and comprehension of language This region encompasses the neural formations that make up the primary motor cortex, which controls the movements involved in speech production and elaboration Consequently, the Broca area also houses the memory for muscle movement sequences, with each sequence linked to a corresponding auditory sequence in the posterior part of the brain.

Teorii care explică recunoaşterea cuvintelor:

The motor theory of speech perception, proposed by Liberman in 1967, suggests that individuals cannot effectively perceive language if each phoneme corresponds solely to a distinct sound Research by Orr, Friedman, and Williams in 1965 demonstrated that subjects could track around 400 words per minute in English, indicating the ability to understand approximately 30 phonemes per second, albeit with some difficulty However, Miller and Taylor's auditory psychophysiological experiments in 1948 revealed that even 15 phonemes per second, a common rate in everyday speech, are not clearly perceived if treated as separate acoustic events This theory posits that listeners "mimic" the necessary movements for producing words as spoken by others Further research supporting this theory found increased tongue movement and heightened activation of the motor cortex controlling tongue movements in listeners when presented with spoken words.

„terra”, decõt ợn cazul prezentării cuvõntului „baffo” (Fadiga, Craighero,

Aceste cercetări nu demonsterază, ợnsă o relaţie cauzală ợntre percepţia limbajului şi activarea cortexului motor, ci doar faptul că facilitează percepţia prin activare.

Acest fapt este demonstrat de cercetările care arată că copiii de 6-8 luni, care nu au bine dezvoltată articularea cuvintelor, se descurcă destul de bine la sarcinile de detecţie a silabelor.

The cohort model theory (Marslen-Wilson and Tyler, 1980) posits that a word is recognized through the gradual reduction of potential candidates as new phonemes emerge Recognition occurs when only one candidate remains, known as the "uniqueness point." This process begins with bottom-up processing, driven by the initial phonemes of the word, where the word is not yet recognized, allowing entry into the lexical system This is followed by top-down processing, where the array of possible words is compared to the original word heard Context plays a crucial role, enabling earlier recognition of words Therefore, a mechanical relationship between the emergence of phonemes and the progression to the uniqueness point cannot be established, as the leap to recognition occurs at this critical juncture.

The revised version of this theory suggests that bottom-up processing does not completely eliminate a candidate word; rather, it partially deactivates it This explanation aims to clarify how we can recognize words even when they are mispronounced from the very beginning.

3 Modelul „TRACE”, al „urmelor” (McClelland şi Elman, 1986;

According to McClelland (1991), the recognition of spoken words involves the simultaneous activation of feature detectors, phoneme detectors, and word detectors This process mirrors the interactive activation model proposed by McClelland and Rumelhart in 1981, which explains the recognition of written words It relies on both bottom-up and top-down processing, as well as the activation and inhibition of lexical units.

According to the theory of interactive activation, excitation and inhibition are transmitted through a network of units that serve as information processing elements, reflecting a connectionist model Each unit acts as a simple processing device, representing a hypothesis about the input being processed The activation of a unit is linked to the strength of the hypothesis it embodies Units with mutually consistent connections are excitatory, while those with inconsistent connections are inhibitory This interactive activation processing allows each hypothesis to both constrain and be constrained by other consistent or inconsistent hypotheses.

Units are organized at various levels, including features, phonemes, and words An inhibitory or activatory action propagates through a network, creating a pattern or "trace." Words are recognized or identified based on the activation level of potential candidate words, demonstrating how words can be recognized even when not pronounced correctly and how they are identified quickly.

Acest model, deşi are multe avantaje, nu poate explica cum ortografia unor cuvinte sau durata unor silabe pentru a fi pronunţate influnţează identificarea cuvintelor.

Understanding language involves two complementary themes: syntax, which focuses on how words are combined, and pragmatics, which pertains to the use and interpretation of language in real-world contexts and intended meanings.

Research focuses on the relationship between syntactic analysis and semantic analysis Grammar should facilitate the formulation of all possible sentences in a language while rejecting any unacceptable ones However, syntactic and grammatical ambiguities may arise, which can be resolved through prosody.

Sunt două tipuri de teorii concurente: teoriile de tip „grădină” şi teoriile bazate pe constrângeri.

Teoriile de tip „grădină” (Frezier şi Rayner, 1982) propun următoarele asumpţii:

1 o singură structură sintactică este luată ợn considerare pentru orice propoziţie;

2 ợnţelesul nu este implicat ợn selecţia structurii iniţiale;

3 cea mai simplă structură sintactică este aleasă pe baza a două principii: ataşarea minimă şi principiul ợnchiderii amõnate.

Principiul ataşării minime: structura gramaticală care produce cele mai puţine noduri (părţi majore ale propoziţiei – verbe şi substantive) este preferată.

Principiul ợnchiderii amõnate: noile cuvinte ợntõlnite ợn propoziţie sunt ataşate frazării curente dacă sunt permise gramatical.

4 dacă cele două principii sunt ợn conflict, atunci conflictul se rezolvă prin alegerea principiului minimei ataşări.

5 dacă structura sintactică construită de o persoană ợn primă fază nu este compatibilă cu ceea ce urmează, atunci ợn faza a doua structura iniţială este revizuită

The theory is validated through experiments that utilize ambiguous sentences while tracking eye movements to identify points of grammatical structure breakdown For instance, in the sentence "I saw the colonel's daughter who was on the balcony," the question of who was on the balcony—either the daughter or the colonel—depends on the placement of the comma, which indicates where the separation occurs If we apply the principle of minimal attachment, we conclude that the daughter was on the balcony Conversely, if we use the principle of final closure, which is not always applied, we deduce that the colonel was the one on the balcony.

Principiile sunt bune pentru că miminizează solicitările către memoria de lucru.

Sunt de inspiraţie conexionistă Susţin că interpretarea depinde de sursele multiple de informaţie (sintactice, semantice, cunoaşterea generală a cuvintelor, etc.) numite

„constrângeri” Ele limitează numărul de interpretări posibile Este susţinută procesarea ợn paralel şi nu succesiv, ca ợn teoria anterioară.

Teoria cea mai influentă este propusă de MacDonald (1994) Presupune că toate tipurile de informaţii sunt disponibile vorbitorului şi sunt activate proceduri de analiză aflate ợn competiţie Metoda este următoarea:

1 cunoştinţele gramaticale limitează din start posibilităţile de interpretare ale propoziţiilor.

2 variatele forme ale informaţiei asociate cu un cuvânt nu sunt independente unele de altele.

3 un cuvõnt poate fi ambiguu ợn anumite feluri şi neambiguu ợn altele (ex ca şi conjugare, dar nu ca şi categorie gramaticală).

4 interpretările gramaticale posibile diferă ợn probabilitate şi frecvenţă de apariţie ợn funcţie de experienţa vorbitorului.

The non-literal meaning transcends the literal interpretation of words, conveying intended messages that consider the current social context and utilize figurative language It is essential to note that non-literal meaning is accessed after the literal meaning, particularly when the former lacks clarity, supporting standard pragmatic theory Kintsch's (2000) non-standard predicative model consists of two components.

1 componenta analizei semantice latente (sensul cuvintelor ợn funcţie de relaţiile dintre ele);

2 componenta constructiv-interpretativă: foloseşte informaţia din prima componentă pentru a construi interpretarea unei declaraţii ợn forma unei structuri de tip

In the context of this theory, an argument is defined as a predicate, which closely aligns with the concepts of "property" or "attribute" in a logical sense rather than a grammatical one The structure F(x) translates to "all x are F," indicating that every instance of x possesses the attribute F For example, in the statement "all girls are beautiful," "beautiful" serves as the logical predicate while "girls" represents the logical subject or the instantiation of the variable x in this case.

Most research does not support the standard theory of pragmatics, indicating that figurative meaning is not accessed automatically but is understood as quickly as literal meaning An experiment by Glucksberg (2003) demonstrated that subjects required more time to differentiate between true figurative sentences and false literal ones, which was unexpected In Kintsch's model, the second component involves selecting relevant features (meanings) of the predicate for the argument while inhibiting irrelevant predicative features For example, in the sentence "The lawyer was like a shark," we infer that the lawyer was aggressive, goal-oriented, and able to dismantle the opponent's arguments This means we selected traits about sharks that fit the context and speaker's intent, while inhibiting irrelevant characteristics like their large teeth or swimming speed.

Memoria

- Să aibă o altă perspectivă asupra memorie;

- Să facă comparaţie ợntre diferitele modele explicative ale memoriei.

3 Eysenck, M W., Keane, M.T (2010) Cognitive Psychology Sixth Edition. Psychology Press

A popular model for explaining memory posits the existence of interconnected storage spaces (levels) that serve different functions for the knowledgeable subject This model suggests that there are three levels of memory through which information flows It is important to note that stating information is stored in memory does not imply it is static; rather, it is in a constant state of rearrangement and processing, making memory both a process and a product.

The sensory memory (SM) serves as the initial stage of information retention, capturing environmental data before it is transferred to short-term memory This process occurs within a brief timeframe of 200-400 milliseconds Information is retained based on sensory modalities, resulting in distinct types of sensory memory, such as visual (iconic) memory and auditory (echoic) memory.

Short-term memory (STM) has a limited capacity of 7±2 chunks, which are meaningful units of information, as highlighted by G Miller in his 1956 study, "The Magical Number Seven, Plus or Minus Two." This type of memory retains information for a duration of 15 to 30 seconds.

The long-term memory (LTM) processes semantic information, possesses unlimited capacity, and can retain information for extended periods In terms of sensory memory, it can be said that there are as many types of memory as there are senses However, the most studied types are iconic memory (visual) and echoic memory (auditory) Iconic memory lasts approximately 200 to 500 milliseconds, is pre-attentional, automatic, and has an unlimited capacity.

Visual stimuli retention is crucial for activating feature detectors, especially for stimuli with very brief exposure times This retention is particularly beneficial during blinks or ocular saccades Information in iconic memory fades quickly, typically within 0.5 seconds (Eysenck & Keane, 2010, p 206).

Echoic memory, also known as auditory sensory memory, lasts longer than iconic memory, retaining stimuli for approximately 2 to 4 seconds (Eysenck & Keane, 2010, p 206) This extended duration is attributed to the anatomy of the auditory analyzer, which requires more time to capture and convert stimuli into internal signals.

Recent research by Jonides et al (2008, as cited in Eysenck & Keane, 2010, p 209) suggests that short-term memory is a temporary activation of representations from long-term memory, indicating a single memory block rather than multiple layers However, this model fails to account for cases where some amnesic patients retain intact short-term memory while struggling with long-term memories Although there is substantial experimental evidence supporting the unitary model, it remains inconclusive which model is superior It is evident that some patients perform poorly on short-term memory tests reliant on long-term memory information, suggesting that short-term memory is indeed a temporary activation of long-term memory data, but it cannot be solely defined as such.

Baddeley and Hitch (1974) proposed a model to refine the concept of short-term memory This model effectively captures both the active processing of information and its storage for a brief duration.

In 1974, the authors initially proposed three components that constitute working memory: the central executive (the central administrator), the phonological loop, and the visuo-spatial scratch-pad The latter two are also referred to as slave systems, as they are subordinate to the central executive.

Fig.10 Modelul memoriei de lucru dupa Baddeley (1986, apud Iordan (2007, p 51)

Ulterior, ợn 2000, Baddeley a revizuit modelul şi a adăugat bufferul episodic.

Acesta este o componentă care face legătura cu memoria de lungă durată şi realizează coerenţa procesărilor din momentul prezent, strõngõndu-le ợntr-o unitate, ợntr-un episod.

Fig 11 Modelul revizuit al memoriei de lucru (după Baddeley, 2000)

Să le luăm pe rând.

Unitatea executivă centrală (eng central executive)

It is a control system that monitors and regulates two subordinate components This system can process both visual and auditory information, making it modality-free Additionally, it is characterized by a limited storage capacity.

Bucla fonologică (eng phonological loop)

Este un sistem subordonat unităţii executive Este formată din două componente:

Sistemul de control articulat (verbal), „vocea” interioară: pentru ca informaţia temporar stocată ợn MSD să treacă ợn MLD trebuie să fie repetată Acest sistem este responsabil pentru această sarcină.

Unitatea de stocare fonologică, „urechea” interioară: permite stocarea temporară a informaţiilor care vor fi preluate şi repetate pentru a fi transferate ợn MLD

Blocnotesul spaţio-vizual (eng visuospatial sketchpad)

Temporary storage and manipulation of spatial and visual information is facilitated by what is known as the "inner eye." This component allows us to visualize information on a mental internal screen Research in this area focuses on the concept of mental imagery.

Figure 12 illustrates working memory in relation to consciousness, highlighting its four components within the conscious realm The central executive is shown to possess intentionality, enabling it to select, abandon, or inhibit the processing of information This indicates a connection between the central executive and deeper processes, such as regulatory functions and moral consciousness The complexity of this component of working memory is noteworthy, as it may complicate the explanation of how working memory operates An overly extensive component could diminish its integral role within working memory.

In the hierarchical model of memory, the Memory Storage Device (MSD) is perceived as a singular unit that temporarily activates information from the Long-term Memory (LTM) while also storing new information Conversely, in the working memory model, the MSD is viewed as comprising multiple components that collaboratively function as a "mental workspace."

Puncte slabe ale modelului memoriei de lucru:

• Rolul unităţii executive centrale este neclar Este cea mai mare slăbiciune a modelului, tocmai datorită importanţei sale pentru model, faptului că această componentă are rolul de coordonare

• Este presupus faptul că unitatea executivă are activităţi de procesare extrem de variate, ceea ce face extrem de dificilă explicarea funcţionării ei

• Este posibil ca unitatea executivă să aibă, la rândul ei, alte componente Care?

• Dacă unitatea executivă are o capacitate redusă, care este aceasta?

• Sunt dovezi care arată că memoria de lucru vizuală este până la un anumit punct separtă de memoria de lucru verbală

• Are foarte multe componente, care sunt toate conştiente Oare toate componenetele şi procesările din memoria de lucru sunt conştiente?

Puncte tari ale modelului memorie de lucru:

• Modelul tratează procesarea activă şi stocarea tranzitorie a informaţiei şi este implicat ợn toate sarcinile cognitive complexe.

• Blocnotesul spaţio-vizual ajută la explicarea orientării geografice.

• Explică efectiv capacitatea noastră de a stoca informaţii temporar, ợn timp ce procesăm acel material.

It is widely accepted that the working memory model depicting the Multiple Store Model (MSD) as composed of several units is more ecologically valid than the traditional hierarchical memory model, which treats MSD as a single unit.

MSD deficiencies in patients with brain injuries can be explained by the fact that these deficiencies are not due to a complete "failure" of memory, but rather to specific components This understanding allows for more effective treatment of these deficiencies.

Memoria de lungă durată (MLD)

Raţionarea şi luarea de decizii

Raţionarea şi luarea de decizii 92

- Să ợnţeleagă perspectiva psihologiei cognitive asupra rezolvării de probleme şi luării deciziilor.

3 Eysenck, M W., Keane, M.T (2010) Cognitive Psychology Sixth Edition. Psychology Press

Raționare vs luare de decizii 1

Se impune o distincție ợntre procesele raționării versus cele ale luării de decizii

Both reasoning and decision-making involve evaluating multiple alternatives and ultimately selecting one According to Hastie, "Decision-making refers to the entire process of choosing a course of action." Reasoning pertains to the components of the broader decision-making process, which includes evaluating, estimating, and deducing events that are likely to occur, as well as the evaluative reactions of the decision-maker regarding these events (Hastie 2001, pg 657).

When analyzing the reasoning process, the focus shifts to how individuals draw conclusions or make deductions regarding abstract situations or events in their surrounding environment.

On the other hand, the decision-making process involves having several alternatives from which the individual must choose one Thus, in decision-making, the list of options is already available, in contrast to reasoning, where the options are generated by the individual who is reasoning.

Another consequence is that reasoning involves finding suitable solutions for problems, while decision-making occurs when options are already available, and estimates are made based on preferences Thus, the primary goal of the reasoning process is to identify effective strategies, whereas in decision-making, the factors analyzed are those that influence the choice.

A significant discrepancy exists between reasoning and decision-making, primarily due to the evaluation criteria used While the accuracy of reasoning is the most frequently assessed characteristic, decision-making focuses on evaluating the consequences of the conclusions reached.

Cercetările efectuate cu privire la procesul raționării

1 Capitol realizat ợmpreună cu stud Cezarina Chirilă (promoţia 2014).

Trei contribuții timpurii aduse acestei dezbateri: empirică, metodologică și teoretică au reprezentat reacția oamenilor de știință la modele descrise mai sus

The empirical contribution proposed by Paul Meehl in 1954 examined the differences between predictions made by clinical experts and probabilities derived from statistical methods His findings revealed that the latter formulas were almost always superior Additionally, his research highlighted a significant discrepancy between clinicians' self-assessments of their performance and their actual success rates.

Ward Edwards made a significant methodological contribution to psychology by introducing Bayesian analysis, establishing a normative standard for everyday reasoning He argued that intuitive reasoning regarding predictions could not yield the same results as the normative standard proposed by the homo economicus model This approach sparked interest in uncovering the causes of suboptimal performance and developing strategies to enhance decision-making processes.

Herbert Simon made a significant theoretical contribution to the field by stating that the "absolute" rationality proposed by the rational decision-making model is an unrealistic standard for human reasoning He introduced the theory of bounded rationality, which highlights the inherent limitations of human cognitive processing While individuals can think rationally and make decisions accordingly, they are constrained by their computational capacities In his 1957 work, Simon also discussed the simplified heuristics that individuals might employ to effectively manage these limitations, linking this approach to complex decision-making.

The article introduces the thinking model proposed by Reverend Thomas Bayes, which later served as a methodological framework for significant research on heuristics and biases Due to its complexity, individuals often do not apply this model in reasoning situations.

„formulele acestuia sunt esențiale pentru evaluarea ipotezelor științifice, pentru obținerea unor diagnostice medicale realiste, pentru analiza datelor demografice și pentru a rezolva multe sarcini din lumea reala” (Eysenck & Keane, 1990, pp 456– 458.).

Bayes' theorem is used to analyze situations with two possible hypotheses (e.g., X is telling the truth vs X is not telling the truth) and demonstrates how new information alters the probability of each hypothesis being true The theorem emphasizes two key components to consider initially: the prior probability (the relative likelihood of the two hypotheses before data is calculated) and the predictive percentage (the relative probability of observing the data under each hypothesis).

Teorema lui Bayes are următoarea formulă:

Kahneman și Tversky au aplicat această teoremă ợn problema taxiurilor (Tversky, A., & Kahneman, D.,1980, p 117-119 și Fishbein M., 1980, p 49-72):

A taxi was involved in an accident one night, and the driver fled the scene In the city, two taxi companies operate: one with green taxis, which make up 85% of the fleet, and another with blue taxis, accounting for 15% In court, a witness identified the fleeing taxi as blue However, the judge assessed the witness's ability to identify taxis under normal visibility conditions When presented with a mix of blue and green taxis, the witness correctly identified the taxi 80% of the time, making mistakes 20% of the time This raises the question of the probability that the taxi involved in the accident was blue rather than green.

Assuming the blue taxi is HA and the green taxi is HB, the initial probability for HA is 0.15 and for HB is 0.85 The probability that the witness correctly identifies the blue taxi is p(D/HA) = 0.80, while the probability of confusing the blue taxi with the green one is p(D/HB) = 0.20 Therefore, we can use these probabilities in our calculations.

The initial probability of the taxi being blue is 12:17, translating to a 41% chance, while there remains a 59% likelihood that it was green Despite these results, which were based on fundamental probability concepts from classical rational models, participants in the experiment decided based on the witness testimony, estimating an 80% probability that the taxi was blue.

Rezultatele acestui studiu a arătat că, adesea, indivizii nu iau ợn calcul informațiile ce reprezintă frecvența de bază (adică frecvența relativă a unui eveniment ợn cadrul unui populații)

One reason for this trend is the overwhelming presence of various stimuli and types of information in our environment Given our limited capacity to process information and the restricted time available for reasoning, we must selectively focus on the information that is most relevant to us from the vast array of data This concept also applies to decision-making, where time constraints further limit our processing abilities in the pursuit of making a choice.

Heuristics are intuitive rules that assist us in reasoning and decision-making processes that are not cognitively demanding, relying on minimal data or cues for quick judgments, thus saving us time and effort However, we are prone to making mistakes when using these mental shortcuts, even though they simplify the cognitive load of decision-making.

Tiêu đề	Psihologie Cognitivă
Người hướng dẫn	Conf. Univ. Dr. Daniela Dumitru
Trường học	Universitatea Titu Maiorescu
Chuyên ngành	Psihologie
Thể loại	modul
Năm xuất bản	2014

Định dạng
Số trang	137
Dung lượng	2,6 MB