Seven-month-old infants chunk items in memory

Although working memory has a highly constrained capacity limit of three or four items, both adults and toddlers can increase the total amount of stored information by "chunking" object representations in memory. To examine the developmental origins of chunking, we used a violation-of-expectation procedure to ask whether 7-month-old infants, whose working memory capacity is still maturing, also can chunk items in memory. In Experiment 1, we found that in the absence of chunking cues, infants failed to remember three identical hidden objects. In Experiments 2 and 3, we found that infants successfully remembered three hidden objects when provided with overlapping spatial and featural chunking cues. In Experiment 4, we found that infants did not chunk when provided with either spatial or featural chunking cues alone. Finally, in Experiment 5, we found that infants also failed to chunk when spatial and featural cues specified different chunks (i.e., were pitted against each other). Taken together, these results suggest that chunking is available before working memory capacity has matured but still may undergo important development over the first year of life.     

工作记忆中“组块”概念的演化及理论模型

在Miller提出“神奇的数字7±2”之后, “块”被很多理论作为个体工作记忆加工过程中具有稳定结构并可用于衡量记忆容量的单位。但随着研究者对“组块”研究的深入, 他们对组块的定义也在发生着改变。与此同时, 不少研究发现个体的年龄阶段与其主要采用的组块层级相对应, 但尚不清楚组块层级的转换是否存在固定的年龄区间, 且对组块机制的解释仍存在分歧。因此本文针对组块定义的发展与演变、年龄阶段特征及其机制三方面展开综合讨论。未来的研究可以更多探讨长时记忆在工作记忆组块运行机制中的作用, 完善不同年龄阶段个体的组块特征, 以及怎样发挥复述策略和“少即是多”原则在组块过程中的优势等问题。     

Why computational models are better than verbal theories: the case of nonword repetition

Tests of nonword repetition (NWR) have often been used to examine children's phonological knowledge and word learning abilities. However, theories of NWR primarily explain performance either in terms of phonological working memory or long‐term knowledge, with little consideration of how these processes interact. One theoretical account that focuses specifically on the interaction between short‐term and long‐term memory is the chunking hypothesis. Chunking occurs because of repeated exposure to meaningful stimulus items, resulting in the items becoming grouped (or chunked); once chunked, the items can be represented in short‐term memory using one chunk rather than one chunk per item. We tested several predictions of the chunking hypothesis by presenting 5–6‐year‐old children with three tests of NWR that were either high, medium, or low in wordlikeness. The results did not show strong support for the chunking hypothesis, suggesting that chunking fails to fully explain children's NWR behavior. However, simulations using a computational implementation of chunking (namely CLASSIC, or Chunking Lexical And Sub‐lexical Sequences In Children) show that, when the linguistic input to 5–6‐year‐old children is estimated in a reasonable way, the children's data are matched across all three NWR tests. These results have three implications for the field: (a) a chunking account can explain key NWR phenomena in 5–6‐year‐old children; (b) tests of chunking accounts require a detailed specification both of the chunking mechanism itself and of the input on which the chunking mechanism operates; and (c) verbal theories emphasizing the role of long‐term knowledge (such as chunking) are not precise enough to make detailed predictions about experimental data, but computational implementations of the theories can bridge the gap.     

Infants chunk object arrays into sets of individuals

Research suggests that, using representations from object-based attention, infants can represent only 3 individuals at a time. For example, infants successfully represent 1, 2, or 3 hidden objects, but fail with 4 (Developmental Science 6 (2003) 568), and a similar limit is seen in adults' tracking of multiple objects (see Cognitive Psychology 38 (1999) 259). In the present experiments we used a manual search procedure to ask whether infants can overcome this limit of 3 by chunking individuals into sets. Experiments 1 and 2 replicate infants' failure to represent a total of 4 objects. We then show that infants can exceed this limit when items are spatiotemporally grouped into two sets of 2 prior to hiding, leading infants to successfully represent a total of 4 objects. Experiment 3 demonstrates that infants tracked the 4 objects as two sets of 2, searching for each set in its correct hiding location. That infants represented the number of individuals in each set is demonstrated by their reaching for the correct number of objects in each location. These results suggest that by binding individuals into sets, infants can increase their representational capacity. This is the first evidence for chunking abilities in infants.     

Models of verbal working memory capacity: what does it take to make them work?

Theories of working memory (WM) capacity limits will be more useful when we know what aspects of performance are governed by the limits and what aspects are governed by other memory mechanisms. Whereas considerable progress has been made on models of WM capacity limits for visual arrays of separate objects, less progress has been made in understanding verbal materials, especially when words are mentally combined to form multiword units or chunks. Toward a more comprehensive theory of capacity limits, we examined models of forced-choice recognition of words within printed lists, using materials designed to produce multiword chunks in memory (e.g., leather brief case). Several simple models were tested against data from a variety of list lengths and potential chunk sizes, with test conditions that only imperfectly elicited the interword associations. According to the most successful model, participants retained about 3 chunks on average in a capacity-limited region of WM, with some chunks being only subsets of the presented associative information (e.g., leather brief case retained with leather as one chunk and brief case as another). The addition to the model of an activated long-term memory component unlimited in capacity was needed. A fixed-capacity limit appears critical to account for immediate verbal recognition and other forms of WM. We advance a model-based approach that allows capacity to be assessed despite other important processing contributions. Starting with a psychological-process model of WM capacity developed to understand visual arrays, we arrive at a more unified and complete model.     

Processing capacity defined by relational complexity: implications for comparative, developmental, and cognitive psychology

Working memory limits are best defined in terms of the complexity of the relations that can be processed in parallel. Complexity is defined as the number of related dimensions or sources of variation. A binary relation has one argument and one source of variation; its argument can be instantiated in only one way at a time. A binary relation has two arguments, two sources of variation, and two instantiations, and so on. Dimensionality is related to the number of chunks, because both attributes on dimensions and chunks are independent units of information of arbitrary size. Studies of working memory limits suggest that there is a soft limit corresponding to the parallel processing of one quaternary relation. More complex concepts are processed by "segmentation" or "conceptual chunking." In segmentation, tasks are broken into components that do not exceed processing capacity and can be processed serially. In conceptual chunking, representations are "collapsed" to reduce their dimensionality and hence their processing load, but at the cost of making some relational information inaccessible. Neural net models of relational representations show that relations with more arguments have a higher computational cost that coincides with experimental findings on higher processing loads in humans. Relational complexity is related to processing load in reasoning and sentence comprehension and can distinguish between the capacities of higher species. The complexity of relations processed by children increases with age. Implications for neural net models and theories of cognition and cognitive development are discussed.     

Investigating the childhood development of working memory using sentences: New evidence for the growth of chunk capacity

Child development is accompanied by a robust increase in immediate memory. This may be due to either an increase in the number of items (chunks) that can be maintained in working memory or an increase in the size of those chunks. We tested these hypotheses by presenting younger and older children (7 and 12 years of age) and adults with different types of lists of auditory sentences: four short sentences, eight short sentences, four long sentences, and four random word lists, each read with a sentence-like intonation. Young children accessed (recalled words from) fewer clauses than did older children or adults, but no age differences were found in the proportion of words recalled from accessed clauses. We argue that the developmental increase in memory span was due to a growing number of chunks present in working memory with little role of chunk size.     

Developing TODAM: Three models for serial-order information

TODAM2, a theory of distributed associative memory, shows how item and associative information can be considered special cases of serial-order information. Consequently, it is important to get the right model for serial-order information. Here, we analyze and compare three distributed-memory models for serial-order information that use TODAM’s convolution-correlation formalism. These models are the chaining model, the chunking model, and a new model, the power-set model. The chaining model associates each item with its predecessor; the chunking model uses multiple convolutions andn-grams to form chunks; and the power-set model interassociates all items in a set in a particular way to form a chunk. The models are compared in terms of their performance on seven basic tests of serial-order information—namely, serial recall, backward recall, recall of missing items, sequential probe tests, positional probe tests, serial-to-paired-associate transfer, and item recognition. The strengths and weaknesses of each model are discussed.     

Something old,something new: a developmental transition from familiarity to novelty preferences with hidden objects

Novelty seeking is viewed as adaptive, and novelty preferences in infancy predict cognitive performance into adulthood. Yet 7‐month‐olds prefer familiar stimuli to novel ones when searching for hidden objects, in contrast to their strong novelty preferences with visible objects ( Shinskey & Munakata, 2005 ). According to a graded representations perspective on object knowledge, infants gradually develop stronger object representations through experience, such that representations of familiar objects can be better maintained, supporting greater search than with novel objects. Object representations should strengthen with further development to allow older infants to shift from familiarity to novelty preferences with hidden objects. The current study tested this prediction by presenting 24 11‐month‐olds with novel and familiar objects that were sometimes visible and sometimes hidden. Unlike 7‐month‐olds, 11‐month‐olds showed novelty preferences with both visible and hidden objects. This developmental shift from familiarity to novelty preference with hidden objects parallels one that infants show months earlier with perceptible stimuli, but the two transitions may reflect different underlying mechanisms. The current findings suggest both change and continuity in the adaptive development of object representations and associated cognitive processes.     

The magical number 4 in short-term memory: a reconsideration of mental storage capacity

Miller (1956) summarized evidence that people can remember about seven chunks in short-term memory (STM) tasks. However, that number was meant more as a rough estimate and a rhetorical device than as a real capacity limit. Others have since suggested that there is a more precise capacity limit, but that it is only three to five chunks. The present target article brings together a wide variety of data on capacity limits suggesting that the smaller capacity limit is real. Capacity limits will be useful in analyses of information processing only if the boundary conditions for observing them can be carefully described. Four basic conditions in which chunks can be identified and capacity limits can accordingly be observed are: (1) when information overload limits chunks to individual stimulus items, (2) when other steps are taken specifically to block the recording of stimulus items into larger chunks, (3) in performance discontinuities caused by the capacity limit, and (4) in various indirect effects of the capacity limit. Under these conditions, rehearsal and long-term memory cannot be used to combine stimulus items into chunks of an unknown size; nor can storage mechanisms that are not capacity-limited, such as sensory memory, allow the capacity-limited storage mechanism to be refilled during recall. A single, central capacity limit averaging about four chunks is implicated along with other, noncapacity-limited sources. The pure STM capacity limit expressed in chunks is distinguished from compound STM limits obtained when the number of separately held chunks is unclear. Reasons why pure capacity estimates fall within a narrow range are discussed and a capacity limit for the focus of attention is proposed.     

Position effects in chunked linear orders

Summary When someone is asked which of two items from a linearly ordered set of items comes first in that order, reaction time decreases with increasing distance between the two items in the given order. This is known as thedistance effect, a very robust finding in studies on linear orders. Surprisingly, investigations onchunked linear orders show unexpected diverse results. For example, reaction times todifferent probes (i. e., probes with items from different chunks) are sometimes longer than those tosame probes (i. e., probes with items from the same chunk). But if subjects' latency depends on the discriminability of chunk labels vs. exact position information within a chunk,different probes should always yield faster order judgments thansame probes. Two experiments investigated whether position effects, including both end-of-list and end-ofchunk effects, can account for the unexpected finding that between-chunk probes are sometimes answered more slowly than within-chunk probes. Moreover, the linea-rorder paradigm was extended to complex action sequences with more than two chunks, using original cooking recipes as materials. The results showed thatchunk-position effects play a crucial role in comparative judgments. Probes from the first and the last chunk were answered faster than those from a middle chunk.Border items (i. e., items next to a chunk boundary) did not produce a significant effect. However, chunking affected subjects' reaction time, as was indicated by the absence of a distance effect fordifferent probes. Besides, action sequences that are ordered in time, as well as concept lists used previously, can be considered linear orders, thus supporting the view that linear orders reflect a basic representation schema of human memory.     

What’s the object of object working memory in infancy? Unraveling ‘what’ and ‘how many’

Infants have a bandwidth-limited object working memory (WM) that can both individuate and identify objects in a scene, (answering ‘how many?’ or ‘what?’, respectively). Studies of infants’ WM for objects have typically looked for limits on either ‘how many’ or ‘what’, yielding different estimates of infant capacity. Infants can keep track of about three individuals (regardless of identity), but appear to be much more limited in the number of specific identities they can recall. Why are the limits on ‘how many’ and ‘what’ different? Are the limits entirely separate, do they interact, or are they simply two different aspects of the same underlying limit?We sought to unravel these limits in a series of experiments which tested 9- and 12-month-olds’ WM for object identities under varying degrees of difficulty. In a violation-of-expectation looking-time task, we hid objects one at a time behind separate screens, and then probed infants’ WM for the shape identity of the penultimate object in the sequence. We manipulated the difficulty of the task by varying both the number of objects in hiding locations and the number of means by which infants could detect a shape change to the probed object. We found that 9-month-olds’ WM for identities was limited by the number of hiding locations: when the probed object was one of two objects hidden (one in each of two locations), 9-month-olds succeeded, and they did so even though they were given only one means to detect the change. However, when the probed object was one of three objects hidden (one in each of three locations), they failed, even when they were given two means to detect the shape change. Twelve-month-olds, by contrast, succeeded at the most difficult task level.Results show that WM for ‘how many’ and for ‘what’ are not entirely separate. Individuated objects are tracked relatively cheaply. Maintaining bindings between indexed objects and identifying featural information incurs a greater attentional/memory cost. This cost reduces with development. We conclude that infant WM supports a small number of featureless object representations that index the current locations of objects. These can have featural information bound to them, but only at substantial cost.     

A memory span of one? Object identification in 6.5-month-old infants

Infants' abilities to identify objects based on their perceptual features develop gradually during the first year and possibly beyond. Earlier we reported [Káldy, Z., & Leslie, A. M. (2003). Identification of objects in 9-month-old infants: Integrating 'what' and 'where' information. Developmental Science, 6, 360-373] that infants at 9 months of age are able to use shape information to identify two objects and follow their spatiotemporal trajectories behind occlusion. On the other hand, another recent study suggests that infants at 4-5 months of age cannot identify objects by features and bind them to locations [Mareschal, D., & Johnson, M. H. (2003). The "what" and "where" of object representations in infancy. Cognition, 88, 259-276]. In the current study, we investigated the developmental steps between these two benchmark ages by testing 6.5-month-old infants. Experiment 1 and 2 adapted the paradigm used in our previous studies with 9-month-olds that involves two objects hidden sequentially behind separate occluders. This technique allows us to address object identification and to examine whether only one or both object identities are being tracked. Results of experiment 1 showed that 6.5-month-old infants could identify at least one of two objects based on shape and experiment 2 found that this ability holds for only one, the last object hidden. We propose that at this age, infants' working memory capacity is limited to one occluded object if there is a second intervening hiding. If their attention is distracted by an intervening object during the memory maintenance period, the memory of the first object identity appears to be lost. Results of experiment 3 supported this hypothesis with a simpler one-screen setup. Finally, results of experiment 4 show that temporal decay of the memory trace (without an intervening hiding) by itself cannot explain the observed pattern of results. Taken together, our findings suggest that at six months of age infants can store but a single object representation with bound shape information, most likely in the ventral stream. The memory span of one may be due to immaturity of the neural structures underlying working memory such that intervening items overwrite the existing storage.     

Factors influencing infants’ ability to update object representations in memory

Remembering persisting objects over occlusion is critical to representing a stable environment. Infants remember hidden objects at multiple locations and can update their representation of a hidden array when an object is added or subtracted. However, the factors influencing these updating abilities have received little systematic exploration. Here we examined the flexibility of infants’ ability to update object representations. We tested 11-month-olds in a looking-time task in which objects were added to or subtracted from two hidden arrays. Across five experiments, infants successfully updated their representations of hidden arrays when the updating occurred successively at one array before beginning at the other. But when updating required alternating between two arrays, infants failed. However, simply connecting the two arrays with a thin strip of foam-core led infants to succeed. Our results suggest that infants’ construal of an event strongly affects their ability to update memory representations of hidden objects. When construing an event as containing multiple updates to the same array, infants succeed, but when construing the event as requiring the revisiting and updating of previously attended arrays, infants fail.     

Reasons for the growth of traditional memory span across age

Three studies, covering together the age range from 3 to 16 years, show the relationships between age, short-term memory span, and language abilities. The growth of the traditional memory span in childhood is systematically related to a concomitant growth of language abilities as tested with CELF-3 Screening, two tests of word knowledge, a test of reception of grammar (TROG-2), and the British Picture Vocabulary Test (BPVS). Linear regression and ANCOVA analyses show that the growth of memory span is fully explained by the growth of language abilities in the three studies that covered different subsamples of children. In a confirmatory factor analysis a two-factor model fits the data well and shows that a “crystallised” language factor is strongly related to age, whereas a “fluid” short-term memory factor is unrelated to age. The two-factor model gives support to Baddeley's multicomponent model of working memory, and the current data suggest that the capacity of the phonological store is fully developed in children younger than 4 years old. The authors hold that chunking capacity, highly dependent on language abilities, is a major source of memory span growth with age. Rather than an increase in the number of chunks, or episodes retained in short-term memory (=chunk span), it is argued that childhood memory span development reflects an increase in the size of each chunk (=chunking capacity).     

The memorial structure of organized sequences

Subjects were asked to memorize a sequence of nine consonants which were grouped into three groups of three letters each (e.g., SBJ FQL ZNG). After learning the sequence, they were presented with single letters, letter pairs, or letter triples and asked to indicate if the probe item appeared in the memorized sequence. The latency results suggest that subjects engage in a linear self-terminating memory search in which the items from one chunk are retrieved from memory as the items from the immediately preceding chunk are being scanned. When the probe consisted of more than one item, the subjects were slowed in their comparison if the letters came from different chunks (e.g., JF vs. FQ in the above illustration), and the number of letters in the probe also influenced the reaction time. Neither of those effects were obtained if all the letters in the probe came from the same chunk, and that would seem to suggest that the probe items from the same chunk were compared in parallel to the letters in the memory item, while items from different chunks were compared serial.     

Recall of random and distorted chess positions: Implications for the theory of expertise

This paper explores the question, important to the theory of expert performance, of the nature and number of chunks that chess experts hold in memory. It examines how memory contents determine players’ abilities to reconstruct (1) positions from games, (2) positions distorted in various ways, and (3) random positions. Comparison of a computer simulation with a human experiment supports the usual estimate that chess Masters store some 50,000 chunks in memory. The observed impairment of recall when positions are modified by mirror image reflection implies that each chunk represents a specific pattern of pieces in a specific location. A good account of the results of the experiments is given by the template theory proposed by Gobet and Simon (in press) as an extension of Chase and Simon’s (1973b) initial chunking proposal, and in agreement with other recent proposals for modification of the chunking theory (Richman, Staszewski, & Simon, 1995) as applied to various recall tasks.     

Memory for multiple visual ensembles in infancy

The number of individual items that can be maintained in working memory is limited. One solution to this problem is to store representations of ensembles that contain summary information about large numbers of items (e.g., the approximate number or cumulative area of a group of many items). Here we explored the developmental origins of ensemble representations by asking whether infants represent ensembles and, if so, how many at one time. We habituated 9-month-old infants to arrays containing 2, 3, or 4 spatially intermixed colored subsets of dots, then asked whether they detected a numerical change to one of the subsets or to the superset of all dots. Experiment Series 1 showed that infants detected a numerical change to 1 of the subsets when the array contained 2 subsets but not 3 or 4 subsets. Experiment Series 2 showed that infants detected a change to the superset of all dots no matter how many subsets were presented. Experiment 3 showed that infants represented both the approximate number and the cumulative surface area of these ensembles. Our results suggest that infants, like adults (Halberda, Sires, & Feigenson, 2006), can store quantitative information about 2 subsets plus the superset: a total of 3 ensembles. This converges with the known limit on the number of individual objects infants and adults can store and suggests that, throughout development, an ensemble functions much like an individual object for working memory.     

What’s magic about magic numbers? Chunking and data compression in short-term memory

Short term memory is famously limited in capacity to Miller's (1956) magic number 7±2-or, in many more recent studies, about 4±1 "chunks" of information. But the definition of "chunk" in this context has never been clear, referring only to a set of items that are treated collectively as a single unit. We propose a new more quantitatively precise conception of chunk derived from the notion of Kolmogorov complexity and compressibility: a chunk is a unit in a maximally compressed code. We present a series of experiments in which we manipulated the compressibility of stimulus sequences by introducing sequential patterns of variable length. Our subjects' measured digit span (raw short term memory capacity) consistently depended on the length of the pattern after compression, that is, the number of distinct sequences it contained. The true limit appears to be about 3 or 4 distinct chunks, consistent with many modern studies, but also equivalent to about 7 uncompressed items of typical compressibility, consistent with Miller's famous magical number.     

Visual working memory declines when more features must be remembered for each object

The article reports three experiments investigating the limits of visual working memory capacity with a single-item probe change detection paradigm. Contrary to previous reports (e.g., Vogel, Woodman, & Luck, Journal of Experimental Psychology. Human Perception and Performance, 27, 92–114, 2001), increasing the number of features to be remembered for each object impaired change detection. The degree of impairment was not modulated by encoding duration, size of change, or the number of different levels on each feature dimension. Therefore, a larger number of features does not merely impair memory precision. The effect is unlikely to be due to encoding limitations, to verbal encoding of features, or to chunk learning of multifeature objects. The robust effect of number of features contradicts the view that the capacity of visual working memory can be described in terms of number of objects regardless of their characteristics. Visual working memory capacity is limited on at least three dimensions: the number of objects, the number of features per object, and the precision of memory for each feature.