Lecture 7: Language, Perception, Attention and Memory

What this lecture covers

The lecture closes out the unit on language by asking whether animals truly have language, and previews the broader questions of language and thought. It then pivots to a new unit that Professor Bloom treats as one continuous story: perception, attention, and memory. The connecting idea is that you see a scene, interpret it, hold it briefly, and later try to retrieve it, and each step can go wrong in informative ways.

By the end you should be able to explain why trained-chimpanzee language claims are weaker than they appear, describe why vision is a much harder computational problem than it looks, and give examples of how attention and memory limit what we notice and recall. This sets up the following lectures on memory failure and suggestibility.

Key ideas

• Animal communication vs. human language: wild animal signaling (vervet monkey calls, bee dances, birdsong) lacks phonology, morphology, syntax and recursion, so calling it "language" in the technical sense is misleading.
• Trained-primate language claims: chimpanzees like Nim Chimpsky produce highly repetitive, non-recursive utterances after years of training, unlike children who acquire words in days.
• Sapir-Whorf hypothesis (linguistic relativity): the idea that the structure of a language you know shapes how you think, distinct from the question of whether language is required for abstract thought at all.
• Perception as inference: the retina delivers a two-dimensional array of neural firings, and the brain must guess the three-dimensional world behind it, since one 2D image is compatible with many possible 3D scenes.
• Unconscious visual assumptions: the visual system corrects for shadows, uses Gestalt cues (proximity, similarity, closure, good continuation, common movement, good form) to group objects, and uses depth cues (binocular disparity, interposition, relative size, texture gradient, linear perspective) to judge distance.
• Visual illusions: illusions like Mueller-Lyer, Ponzo, and Shepard's tables arise because depth cues trigger automatic size-correction that misfires on a flat image.
• Attention as a spotlight: attention is selective; some things (like Stroop word-color conflict) intrude automatically, and divided attention causes people to miss large events, as in the counting-passes demonstration.
• Change blindness: people are often unaware of major changes between two nearly identical scenes, or even when the person they are talking to is swapped for someone else.

Walkthrough

Wrapping up nonhuman communication (0:00)

The lecture opens with two follow-ups from the previous class, then returns to language. Bloom explains that animal communication systems generally fall into three types: finite call lists (vervet monkey alarm calls), continuous analog signals (bee dance encoding food-source richness through intensity), and randomly varied themes (birdsong). None of these show the combinatorial, recursive structure of human language. He is more skeptical than the assigned textbook about famous trained-primate cases, arguing that their utterances are repetitive and lack recursion, and that expecting a chimpanzee to learn human language is like expecting one species to have another's specialized ability, such as bat echolocation.

Language and thought, questions from class (9:16)

Bloom frames two separable questions: whether language is necessary for abstract thought (tested by looking at how smart language-less creatures like infants and chimpanzees are) and whether the specific structure of a known language shapes thinking, the Sapir-Whorf hypothesis. He notes this connects to an assigned reading response. He then takes several student questions on individual differences in language learning, covering the small female advantage, strong genetic influence, bilingual children learning both languages without apparent cost, and hemispheric specialization for language typically on the left side in right-handed people.

Why perception is a hard problem (18:06)

Bloom introduces the perception-attention-memory unit with the claim that perceiving a scene, holding it in memory, and later recalling it are one connected process, each prone to errors people rarely notice in themselves. He argues that vision only looks easy; the retina is a two-dimensional array of neural firings, and reconstructing a three-dimensional world from it is mathematically underdetermined, since many 3D scenes can produce the same 2D image. He then works through three examples of unconscious inference: brightness constancy (a well-known checker-shadow style illusion shows two patches judged different in brightness are physically identical, because the brain factors out apparent shadow), object segmentation using Gestalt principles like proximity, similarity, closure, good continuation, common movement and good form (illustrated with illusory Kanizsa-style triangles and squares), and depth perception using cues such as binocular disparity, interposition, relative size, texture gradient and linear perspective.

Depth cues and classic illusions (1086)

Continuing the depth discussion, Bloom explains the Mueller-Lyer and Ponzo illusions as cases where cues suggesting one line is farther away trigger automatic size correction, making it look larger even though the two lines are equal. He extends this to Roger Shepard's "table" illusion, where two identically shaped tabletops look markedly different in proportion because of the same depth-correction mechanism, and invites students to verify it by tracing one shape onto the other.

Attention as a limited spotlight (39:15)

The lecture shifts to attention and memory together, opening with a staged amnesia demonstration to show that even "losing your memory" leaves language and motor skills intact. Bloom distinguishes several memory types: sensory memory (brief afterimages and echoic memory), short-term/working memory, and long-term memory; explicit versus implicit memory; semantic versus episodic memory; and encoding, storage and retrieval, including recall versus recognition. He then demonstrates attention directly: an automatic pop-out search for a colored letter, a harder search among similar distractors, and the Stroop effect, where naming ink colors is slowed by conflicting color words because reading is too automatic to suppress.

Inattentional blindness and change blindness (51:52)

Using the well-known basketball passing-count video, Bloom shows that roughly half the class misses an obvious unexpected event while focused on counting passes. He then demonstrates change blindness with flicker images and a scene-detail video, and describes Dan Simons's real-world "door" study, in which a person giving directions often fails to notice that the person they are talking to has been swapped for someone else mid-conversation, unless the swap changes the person's apparent sex or race. He closes by taking a few final questions on attention and on the memory case introduced earlier.

Before you watch

• Review the earlier language lectures on phonology, morphology, syntax and recursion, since the animal-communication discussion depends on that vocabulary.
• Have a rough sense of what a neuron's "firing" pattern means, since the perception discussion treats the retina as an array of firing/non-firing signals.
• Come ready to actively watch the demonstration clips (color-word naming, basketball passing, flicker images) rather than reading a description of them, since the lecture's core points depend on experiencing the illusions and blindness effects yourself.

Check your understanding

1. Why does Professor Bloom think trained-chimpanzee language studies are less impressive than they first appear?
2. Explain why reconstructing a 3D scene from a 2D retinal image is described as a mathematically underdetermined problem.
3. Name two Gestalt principles used for grouping objects and give an example of each from the lecture.
4. Why does the Stroop effect happen, and what does it reveal about the automatic nature of some attention?
5. What is change blindness, and how did the "person swap" study demonstrate it outside the laboratory?