The experimental tradition focused primarily on the more dynamic on-line aspects of mental functioning. The aim was to explain how humans: (i) perceive the world and choose information that is relevant at a given moment; (ii) make sense of or understand the information perceived; (iii) solve the problems encountered; and (iv) store and organize their knowledge and experience about the world.
Information processing models have dominated cognitive psychology since the early 1950s. According to these models, humans usually operate under conditions of uncertainty caused by redundant, conflicting, or incongruent information relative to a goal. It is a truism that always we see much more than what is needed at a given moment. For instance, when looking for someone in an open space we often see many other persons or objects around. To find the person we are looking for requires searching for a particular face and bodily outline, quickly rejecting others that may look alike. Also, to make a final choice, we often need to fill in gaps of information, as when the person we are looking for is partially occluded by somebody else. In this case we match what we see with what we remember by drawing on memory or inference. Likewise, when in a conversation people often need to make decisions about the meaning of what was heard or the intentions of their partners. Often there may be conflicting information that must be interpreted, such as when one has the feeling that the partner means something other than what she or he says. To meet their goals, humans must be able to focus attention and process goal-relevant information efficiently, filtering out goal-irrelevant information within the constraints of the fast flow of a conversation and filling in gaps by inference. In effect, controlled attention, speed of processing, working memory, and inference are considered important in registering information, understanding, learning, and problem-solving.
Mechanisms of attention and inhibition
Controlled attention includes processes enabling individuals to stay focused on information of interest while filtering out irrelevant information (MacLeod, 1991; Neill, Valdes, & Terry, 1995). For instance, attention drives the search for particular characteristics when we try to find a person in the crowd. In laboratory situations the Stroop phenomenon is the paradigmatic example of the conditions requiring efficient handling of conflicting information. In the classic version of the task, participants are presented with cards showing colour words printed either in black ink or in an ink colour which is different from the colour denoted by the word itself (for example, the word âredâ printed in blue ink). The participants may be examined under several conditions. Of primary interest is the condition of having to read the colour words which are printed in black ink and the condition of having to name the ink colour of words where meaning and ink colour differ (for instance, say blue when the word is âredâ). Under these conditions, word reading is much faster (43.30 seconds for a card with 100 words) than naming the ink colour (110.3 seconds). The time difference between these two conditions has been ascribed to the interference of the dominant aspect of the stimuli (i.e., the tendency to read a word) with the processing of the weaker but goal-relevant aspect (i.e., the naming of the ink colour). This difference is taken as a measure of inhibition, which is the basic component of controlled attention (MacLeod, 1991; Stroop, 1935).
Speed of information processing was always important in theories of information processing. This is because we usually operate under conditions whereby the information available changes quickly. Thus it is important to be able to perceive and recognize a particular stimulus before it disappears or is overtaken by other stimuli. Speed of information processing basically refers to the time needed to recognize a stimulus or execute a relevant mental operation. Usually, in tests of speed of information processing, the individual is asked to recognize a simple stimulus as quickly as possible, such as a letter, a geometrical figure, or a word in oneâs native language. Under these conditions, speed of processing indicates the time needed to record and give meaning to information. Traditionally the faster an individual can recognize a stimulus the more efficient this individual is considered (Jensen, 1998, 2006; Posner & Raichle, 1997; MacLeod, 1991). A fast speed is considered good for problem-solving because decisions are often made under situations changing rapidly; an understanding of current information must be completed before it is flooded by incoming information.
Mechanisms of representation and processing
It was noted above that understanding requires connecting current information with information already possessed. This requires holding current information active for as long as required to process relations with relevant knowledge from the past. Working memory is the bridge between perception and our knowledge store, or between concepts in the knowledge store (see Figure 1.1). Working memory enables a person to hold information in an active state while integrating it with other information until the current problem is solved. A common measure of working memory is the maximum amount of information and mental operations that the mind can efficiently activate simultaneously. For instance, remember the second last word of each of several sentences, remember the sum of several arithmetic operations, remember where an object was located in a succession of scenes, etc. There is extensive evidence that understanding, learning, and problem-solving are positively related to the capacity of working memory. The assumption is that enhanced working memory increases the connections that can be built between bits of the newly encountered information or between this information and information already stored in long-term memory. Thus enhanced working memory capacity enables us to consider more options in understanding a concept, construct a new concept, or invent solutions to problems.
Figure 1.1 Working memory (WM) mediates between ongoing perception and past knowledge (long-term memory, LTM) on the one hand, and behaviour on the other
Figure 1.2 Baddeleyâs model of working memory. The shaded area represents crystallized systems while the white area represents fluid systems
Baddeleyâs (1990, 2000, 2012) model, which received extensive empirical and theoretical scrutiny, is widely regarded as a good approximation to the architecture of working memory (see Figure 1.2). It posits that working memory consists of a central executive, two specialized storage systems, and an integrative episodic buffer. The central executive is an attentional control system monitoring and coordinating the operation of the two slave systems and coordinating information in working memory with information in long-term memory.
The phonological loop involves a short-term phonological buffer and a subvocal rehearsal loop. The first stores verbal information as encountered; information in this buffer decays rapidly. The second counteracts this decay by refreshing memory traces through rehearsal. The faster rehearsal is, the more the information that can be held in the phonological loop. The visuo-spatial sketchpad is responsible for the retention and manipulation of visual or spatial information. The two slave systems draw on partially different resources. As a result, each is amenable to interference from system-specific information that does not affect the other system. That is, the phonological loop is affected by interference from verbal but not visuo-spatial information; the visuo-spatial sketchpad is affected by visuo-spatial but not verbal information (Shah & Miyake, 1996). However, these systems are interrelated and information from one can be translated into the code of the other through rehearsal guided by the central executive.
The episodic buffer is âa limited-capacity temporary storage system that is capable of intergrading information from a variety of sourcesâ (Baddeley, 2000, p. 421) into unitary multi-dimensional representations using a multi-modal code. It integrates information from the other working memory components and the long-term memory into more complex structures, such as scenes or episodes. It serves as a mediator between subsystems with different codes, such as words, visual images, or number digits. The limited capacity of the central executive affects the integration and maintenance of information within the episodic buffer. The process of retrieving and binding information from multiple sources and modalities is primarily based on conscious awareness.
Baddeleyâs model allows for both specificity and generality in cognitive functioning. Specificity is defined in terms of the modality in which information is received (that is, acoustic or visual) and the ensuing symbol systems, which handle information presented in these modalities (that is, language versus mental imagery). Generality is ensured by the episodic buffer and the central executive. The episodic buffer ensures the communication and production of integrated mental products; the capacity of the central executive sets the general constraints under which the two slave systems can function. Millerâs (1956) famous paper suggested that the capacity of working memory of the normal human adult is 7 units of information plus or minus 2. Later the capacity of working memory was considered to be lower, between 3â5 units, with the difference between the two figures related to capacity needed for the operation of executive processes (Cowan, 2010). We will see below that these aspects of working memory influence intelligence: differences between individuals in working memory are related to differences in their ability to integrate information and reason with it. They also relate to cognitive development, as the capacity of working memory increases with age.
Mechanisms of integration: association, inference, and reasoning
There are several mechanisms of information integration. Association is a more or less automatic mechanism that associates stimuli or responses on the basis of their physical proximity in time or space or their sharing of common characteristics. Various types of learning, such as Pavlovian classical conditioning and Skinnerian operant conditioning, are based on association. In Pavlovian classical conditioning a particular physical stimulus, a bell ringing, takes the properties of another stimulus, food, and then causes the response naturally evoked by this other stimulus, such as salivation at the sight of food, because it occurred just before food appeared. In Skinnerian operant conditioning, a stimulus is associated with a response if the latter appears soon after the former, as when a particular response is rewarded and learned or punished and avoided.
Integration by association may be the basis for inference, because it provides the raw materials for it. Additionally, inference is a more self-directed form of association where links between stimuli are established on the basis of rules encoding what is wrong and right based on past experience. Generally speaking, inference comprises processes that enable thinkers to transfer meaning from one representation to another. This transfer normally occurs on the basis of properties which are present in both the initial (or base) representation and the target representation.
Reasoning is thinking that involves inference. In reasoning, common properties are used as an intermediary between two representations; such that properties characterizing the base representation (apart from common properties) are also ascribed to the target representation. For instance, a four-year-old child thoughtfully concludes: âIf it doesnât break when I drop it, itâs a rock⌠. It didnât break. It must be a rockâ (DeLoache, Miller, & Pierroutsakos, 1998). The two things to be connected here are the notion of the rock and the present object, their common property being that they did not break when dropped. By virtue of this common property, the property of ârockynessâ and other derivative properties (such as rocks are heavy, hard, etc.) may also be transferred to the present object.
There are several types of reasoning, the two most inclusive types being inductive and deductive. Inductive reasoning is a freer kind of reasoning which can involve any kind of representation, such as perceptions, mental images, and propositions of oneâs language. Moreover, in inductive reasoning inference goes from the particular to the general or from the particular to the particular and the conclusion is not necessary but only probable. An example is: âThe dogs I met bark; therefore all dogs bark.â Obviously this conclusion is only probable because in the future we may meet dogs who do not bark. Analogical reasoning is inductive reasoning applied on relationships rather than similarity between the objects themselves. The classic structure of problems which require analogical reasoning is represented by the following formula: a: b: : c: d. For instance, Athens is for Greece what London is for the UK (Holland, Holyoak, Nisbett, & Thagard, 1989).
Deductive reasoning is a more constrained kind of reasoning. First, it only involves verbal statements or propositions, or other symbols which stand for propositions. Thus inference in deductive reasoning is the process which transfers meaning from one set of propositions (the premises) to other propositions (the conclusion). Second, in deductive reasoning the inferential process always proceeds from the general (the premises) to the specific (the conclusion), and the conclusion follows necessarily from the premises. This occurs because in deductive reasoning the premises must be accepted as given, thus the conclusion is mandatory. An example is given below:
Dogs bark
Max is a dog
________________________
Therefore, Max barks.
Obviously, once both premises of this argument are accepted as true the conclusion is necessarily true, no matter what kind of animal Max is or what we know about him. This is more clearly apparent in an argument where the premises are not consistent with reality:
Dogs fly
Max is a dog
________________________
Max flies
Obviously, in this example we have to accept that Max flies, given the premises, even if we know that dogs do not fly.
Both inductive and deductive reasoning implicate a number of different varieties, each of which comprises several inferential processes. For example, inductive reasoning involves statistical reasoning, which focuses on probabilities and analogical reasoning, which focuses on relational similarity. Deductive reasoning involves categorical reasoning and conditional reasoning. Categorical reasoning is based on class relations. For instance, if a specific property characterizes a class (e.g., dogs bark) then this property necessarily goes to the members of the class (all sorts of dogs). Conditional reasoning examines the relationships between âif ⌠thenâ types of propositions. Conditional reasoning is important for intelligent functioning because it allows integration and evaluation of information (Johnson-Laird & Khemlani, 2014).
Conditional reasoning is grounded on four logical schemes, slowly mastered throughout childhood and adolescence (Markovits & Vachon, 1990; Moshman, 2011; MĂźller, Overton, & Reene, 2001): modus ponens (MP), modus tollens (MT), affirming the consequent (AC) and denying the antecedent (DA). Two of the schemes, MP and MT, are decidable and rather easy to grasp because all information needed for a conclusion is present in the premises. In MP, if one accepts that âif A then Bâ and âA occursâ, one must also accept that âB necessarily occursâ. In MT, if B did not occur it necessarily follows that A did not occur. The remaining two, AC and DA, are not decidable because the conclusion depends on information not given in the premises. Specifically, in AC, if B occurs it does not follow that A would also occur, because a third, non-specified factor may be involved. In DA, it does not follow that B would not occur if A does not occur, because a third factor may cause B. Thus these two schemes are called âlogical fallaciesâ because they may deceive the thinker into drawing a conclusion that is not tenable. We will see in the chapters following that MP and MT are attained early in development, at 7â9 years, by practically everyone. The two fallacies are not mastered before the age of 11â12 years and then no more than about one third of adults can handle them systematically (Gauffroy & Barrouillet, 2009; Johnson-Laird & Wason, 1970; Markovits, 2014; Moshman, 2011; Overton, 1990; Ricco, 2010; Wason & Evans, 1975).
Figure 1.3 Wasonâs selection task examining conditional reasoning
Wasonâs selection task is a famous problem demonstrating these processes. This task involves four cards marked with a letter on one side and a number on the other. For instance, the four cards have an A, D, 3, and 7, respectively, as shown in Figure 1.3. The participants were told that the âfollowing rule applies to the four cards and may be true or false: If there is an A on the one side of the card, then there is a 3 on the other side of the card.â They were then asked to indicate which cards must be turned over to decide if the rule is true or false. The correct answer is A and 7. A is relevant because it is stated in the rule. If there is anything but three on the other side the rule would be proved false. Obviously, this is a test of the MP argument. D is irrelevant because the rule does not refer to cards with letters other than A. The card with 3 appears relevant but it is not because the rule specifies what must follow if A occurs and does not state what must follow if 3 occurs. Thus even if an A does not appear on the other side of the â3â card the rule is not contradicted, because the rule does not state what should be marked on the back side of 3. This is the affirmation of the consequent (i.e., the AC fallacy), as explained above. Finally, the card with 7 is relevant because it might have an A on the other side. This is not permitted by the rule, because if there is an A there should be a 3. Thus turning th...
